AT MICROFICHE REFERENCE Lll3RARY A project of Volunteers in Asia Small-Scale Sblar-Powered Irrigation Technical and tconomic Revrew by Sir William Halcrow and partners Pumping Systems: in association with ITDG Pub1ished by: World Bank 1818 H. Street, N.W. WashingtonD.C. 20433 USA Aiailable from: same as above Reproduced by permission. Reproduction of this microfiche document form is subject to the same restrictions of the original document. in any as those UNDP Project CLO/78/@04 Executed by The World Bank SMALL- SCALE SOLAR- POWERED f RRIGATION PUMPING SYSTEMS TECHNICAL AND ECONOMIC REVIEW Sir William Halcrow and Partners in association with the IntermediateTechnology September1981 Development Croup Ltd. London or 8 Hitwally World Sank 1818 8 street Washington CC 20433 USA I have pleasure in enclosing ten copies of the second part of our Final Report on Phase I of the UNDP Project under the title "Small Scale Solar . Powered Irrigation Pumping Systems: Technical and Economic Review". This document supersedes the State of Art Report first submitted to you under cover of our letter dated 21 January 1980. The final text is hased closely upa" a draft prepared for and circulated at the UNDP/Korld Sank Workshop on solar Pumping in Developing Countries held in Manila, Philippines in June 1981. It has been extensively edited since thentoeake account of points made at the Workshop, in discussion with you.?selves and your colleagues at the Sank. and our own advisors. As agreed with you and Mr Dosik the Review includes an extensive section on System Economics which in view of its importance has been placed in a separate chapter in the final text. under the title "Economic and Technic+1 feasibility". For the convenience of readers the Executive Report is included as an Appendix.. I believe that this Technical and Economic Review will provide a valuable for all those concerned with solar powered water pumping and trust satisfactorily fulfLls your requirements. reference it Summary of the Phase 1 Project Yours sincerely A n M”iI wood TECEIN~~~~ANDECON~MICRE~E~ This Volume was prepared during the UNDP Project GL0/78/004 to test and demonstrate suitable small-scale solar-powered pumping systems. It reviews the use of solar pumps for the irrigation of crops on small land-holdings in developing countries and examines the technical and economic criteria which have to be satisfied if this pumping technology is to be adopted. This Volume supersedes the State of Art Report completed in December 1979 and submitted to the World Bank on 21 January 1980 (Ref. 1). A companion Volume “Small-Scale Solar-Powered Irrigation Pumping Systems Phase I Project Report” summarises the work undertaken from July 1979 to May 1981 on field trials, laboratory tests and system design studies which were carried out as part of the UNDP Project. For the :ce of readers the Executive Summary come of the .roject Report is included in the Review as Appendix 3. Both Volumes are available from the World Bank. NOTICE This report was prepared as part of a pmject financed by the UNITED NATIONS DEVELOPMENT PROGRAMME and executed by the WORLD BANK. Neither the UNDP nor the WORLD BANK makes any warranty, express or implied, or assumesany legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial products, process, or service by trade-name, mark, manufacture, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favouring by either the DND? or the World Bank. The views and opinions of authors as expressed herein do not necessarily state or reflect those of the IJNDP or WORLD BANK. / ,‘~ ,,./“ CONTENTS / Jzxeeutivesummary 1. Introduction i.l 1.2 1.3 1.4 i 2. ,.j’ / I” Background to Review / UNDP/World Bank Pro3’ct Context of Review Scope of Review ./.// Role of SmaII-ScaleSolar Pumping for Irrigation 2.1 2.2 2.3 2.4 2.5 The Increasing Importance of Irrigation The “Energy Crisis” and Irrigation Small-Scale Irrigation Power Requirements for Irrigation Pumping Methods Available The Suitability of Solar Pumps for Irrigation Size and Efficiency Considerations for Solar Pumps Altematice Applications for Solar Pumps PageNo. (9 1 : 2 3 4 4 4 6 7 12 16 17 26 .~ 2.6 2.7 2.8 3. Economic and Technical Feasibility 30 3.l 30 30 50 55 3.2 3.3 3.4 4. Introduction System Economics Technical Requirements The Importance of Local Manufacture Solar Pumping Technology - Photovoltaic Systems 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 Photovoltalc Cells Existing Photovoltaic Pumping Installations Photovoltaic Pumping Systems Photovoltaic Arrays Electric Motors j. Batteries Pumps Mechanical Transmissions Photovoltaic Pumping System Optimisation Power Conditioners and Maximum Power Point Trackers The Project Design Study Mathematical Model Results obtained from Model Testing 58 58 64 64 67 78 80 81 90 91 95 99 103 PageNo. 5. Solar 5.1 5.2 53 ;:4 5.5 5.6 5:t 5.8 5.9 6. History Existing Solar Thermal Pumping Installations Thermal Efficiency of Heat Engines Solar Thermal Collectors Rankine Cycle Engines Stirling Cycle Engines Transmissions and Pumps Laboratory Testing of Thermal Systems Thermal System Design Studies Other Solar Pumping System Options 6.1 6.2 6.3 6.4 6.5 6.6 6.1 6.8 7. Pumping Technology - Thermal Systems Introduction Thermoelectric Generators Thermionic Generators Rrayton (Gas Turbine) Solar Thermal Systems Photochemical Systems Improved Efficiency Photovoltaic Technology Memory Metal Heat Engine Osmotic PressureEngines References 110 110 113 113 114 118 124 127 127 130 145 145 145 146 146 146 146 147 147 149 Appendices 1. 2. 3. 4. Preliminvy estimates of costs of solar pumping systemsin developing Countries General Recommendations for the development of Small-Scale Solar Pumping Systems Executive Summary of Project Report Objectives of and Preparation for Phase II of the Project Al A7 A15 A31 LIST OF TABLES Title Tabk 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. II. 12. Irrigated Areas of the World in 1972 Irrigation Water Demand and Solar Energy Availability for Cotton-Wheat Cropping Pattern (Lake Chad Region) Comparison of Principal Methods of Irrigation Pumping Parameters for Baseline Model Comparison Between Computed Costs of Solar and Engine Pumps (Baseline Model) Results of Sensitivity Analysis SmalI-Scale Solar Pumping Installations Results of making Improvements to a PV Pumping System by using the Mathematical Simulation Model Large-Scale Sofretes Solar Thermal Pumping Installations Solar Collector Types used in Thermal Design Studies Example of Thermal System Costing Results of Thermal System Mathematical Modelling Appendix 3 incIudes the foIIwing I II III Iv V tables: Field trials - solar systems costs and data collected Summary of system field performances Laboratory tested systems and components Sensitivity analysis for pumping systems Results of making an Improvement for pumping system by using the mathematical simulation model PageNo. 5 IO 13 32 36 38 61 -63 109 111 133 134 _ 137 LIST OF FIGURES Fisure 1. 7“. 3. 4A 4B 5A 5B 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 21. 28. 29. 30. 31. 32. 33. 34. 35. 36. 31. 38. litle Solar Energy Availability and Crop Irrigation Water Demand (Lake Chad region) Feasible Options for Solar-Powered Pumping Systems Power Output Requirements for Various Heads and Delivery Rates Variation of Irradiance Level Received by a Fixed Collector at Optimum Inclination Variation of Irradiance Level Received by a Sun-Tracking Collector Variation of Irradiance Level Received by Collector Repositioned Once per Day Variation of B-radiance Level Received by Collector Repositioned Twice per Day Losses in a Typical Solar Photovoltaic Pumping System Lossesin a Typical Solar Thermal Pumping System Effects of Inflation and Discount Rate on Annual Cash Flows 1981-2000 Results of Sensitivity Analysis on Solar Pumps Effects of Pumping Head on Water Unit Costs Effects of Water Demand on Water Unit Costs Photovoltaic Module/Array Price Goals and History (in 1980 S) Schematic Arrangement of a Photovoltaic Solar Pumping System Silicon Solar Cell Voltage -Current and Voltage-Power Characteristics of a Silicon S$ar Cell Dependence of Efficiency, I,, and V,, on CeUTemperature Effect of Ceil Temperature on V-I Characteristic Effect of Change in Irradiance on V-I Characteristic Cadmium Sulphide Solar Cell Shottky Barrier (MIS) Solar Cell Gallium Arsenide Solar CeU Methods of Concentrating Sunlight on Photovoltaic Cells Typical dc Permanent Magnet Motor Performance Typical Centrifugal Pump Performance Centrifugal Pump Performance with Flat Speed Characteristics Regenerative Centrifugal Pump Performance Typical Positive Displacement Pump Performance Typical Rotary Positive Displacement Pump Performance Free Diaphragm Pump Performance Performance Characteristics of Photovoltaic System Components Performance of Pompes Guinard System in Mali v Irradiance Daily Output of Pompes Guinard System in Mali Performance of Arco Solar System in Sudan v Irradiance Daily Output of Arco Solar System in Sudan Block Diagram for Photovoltalc System Model Validation of Photovoltaic System Model (Pump Output) Variation of Daily Overall System Efficiency with Head for Array Optimised System Variation of Output with Head for Photovoltaic Systems PageNo. 11 15 19 22 22 24 24 27 :: 41 43 45 59 65 66 66 68 68 68 73 73 73 75 79 82 83 84 a5 86 87 92 94 96 97 98 100 101 104 105 Title 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. Variation of Daily Overall System Eftlciency with Head for System with Maximum Power Point Tracker Effect of Pipework Changeson Specific Capital Cost Comparison of Theoretical Carnot Efficiency with Efficiencies Obtained in Practice (T, - 25O) Schematic Arrangement of a Thermal System Comparison of Solar Collector Performances Simple Rankine Cycle Rankine Cycle with Intermediate Heat Exchanger Hindustan - Brown-Boveri Mark 1 Liquid Piston Rankine Cycle System “Camel” Gravity Operated System Principle of Fhddyne Pump General Arrangement of Sunpower Inc. Stirling Engine Schematic of Solar Pump Corporation Solar Pump Examples of Solar CoUector and Engine Efficiency Effect of Optimum Operating Temperature on Costs of Thermal systema Effect of Optimum Operating Temperature on Collector Areas of Thermal Systems PageNo. 10-l 108 !I2 115 117 119 119 121 123 I26 I28 129 139 140 141 (8 SMALL-SCALE SOLAR-POWERED PUMPING SYSTEMS TECHNICAL AND ECONOMIC REVIEW EXECUTIVE SUMMARY This report reviews the use of small-scale solar powered pumping systems for the irrigation of crops in small land-holdings in developing countries (i.e. of the order of 1 ha). The introductory chapter places this Review in the context of the wider UNDP funded project GL0/78j004, which also involved practical testing of systems. This is followed by a discussion of the principal prime-mover power options available for smali-scale irrigation pumping and how they compare, followed by general discussion of the appropriate sizing of systems, engineering requirements (to suit the operational environment) and the importance of local manufacture of systems within developing countries. The main body of this Review then consists of a technical assessmentand appraisal of the principal types of solar pumping system available and the apparent merits and demerits of numerious subsystem and component choices. A brief historical outline of the development of these technologies is included, with information on currently operational systems and .some assessmentsof possible future trends. This technical assessmentand appraisal is subdivided broadly into three sections. dealing with (i) photovoltaic, (ii) thermal and (iii) unconventional systems that may have a future role. Finally. there is a brief section which attempts to compare the relative eftkiencies and hence the relative costs of the subsystems that comprise the principal small-scale solar pumping system options that we currently available. This section is necessarily speculative, being based on the very tentative cost data associated with today’s immature technology in this field, but it does indicate that it should be possible to achieve acceptably low capital costs for future systems once technical maturity and full scale production are achieved. I 1 XNTRODUCTION 1.1 Background to Review The direct use of solar power may in future find widespread application by farmers in developing countries for the small-scale pumping of irrigation water. The technical feasibility of solar powered pumping has been demonstrated using several different methods of energy conversion, but up to the present it has generally appeared that the technology is too expensive ro be economically viable when compared with conventional alternatives, such as diesel or mains electric pumps. Furthermore, the equipment is often not sufficiently simple and robust to be appropriate for use and upkeep by farmers in developing countries nor has it yet been developed to the stage of being a mature product. With few exceptions, ail the solar pumping equipment available at present is of prototype status, few models having been manufactured in any quantity. The present Project has been specifically restricted to the testing and demonstration of small-sca!esolar-powered pumping systems capable of providing a flow of not less than one litre per second, primarily for irrigation purposes. To be economically attractive, the World Bank/UNDP considered that the pumping systems would ultimately need to deliver water at a cost not exceeding US 8 0.05 per cubic metre (I 979 prices). As the power requirement and hence the unit cost of water pumped increasesin direct proportion to the total head against which the water is pumped (for a given flow), a point must obviously be reached at which it becomes uneconomic to pump through greater heads. The combination of head and flow at which pumping becomes uneconomic is site specitic and depends on costs of the pump and maintenance, crop water requirement and the extra income expected to accrue from improved irrigation. No universally applicable value for this limiting head can be cited, but it is almost certain to be less than 1Om. It was the informal but considered judgement of irrigation advisers to the World Bank and othea that for the purposes of this Project attention should be concentrated on pumps in the hydraulic power output* range of 100 to 500 watts, although systems up to 2000 watts should not be excluded. Considerably larger power outputs are technically possible, but the size ra;rrl?ion adopted for this Project is appropriate to the needs of many millions of family farms and small holdings in the developing world. In particular, the small-scale approach keeps the capital cost down and avoids the problems of multiple uses, with the associated costs of water distribution and control. The importance of the small-scale approach is fully discussed in the Project Report and in this Review, and endorsed by many other references, (e.g. Ref. 2). 1.2 UNDPjWorld Bank Project As stated in the Project Document signed by the World Bank and UNDP in June 1978, this Project forms part of an overall search to develop small-scale pumping systems for water supply and irrigation applications in developing countries which: * ‘hydraulic power output’ means pumped water output calculated on the basis of the product of flow and tofal (pumped head). 2 C) d) e) are based on renewable energy sources; are decentralized; have costs low enough for small farmers; have minimal and simple operation and maintenance requirements; and have good prospects for local manufacture and/or assembly. The UNDP and World Bank decided that the work should first concentrate on the use of solar energy and investigate its application to irrigation pumping. The first phase of the Project was mounted with the overall objective of advising the UNDP and World Bank on whether solar pumping technology was in a position such that it would be worth promoting its development to make it appropriate for pumping water under the conditions that prevail on small farms in the developing world and, if so, what steps should be taken. The enquiry was thus open, although it was expected that the potential of the technology would be recognised and that further dtrelopment would be recommended. The main activities in PhaseI included field trials of possible systems, laboratory tests on principal components and system design studies. In undertaking this work the importance of the potential manufacture (or at least assembly) of systems in developing countries themselves was recognised. At a very early stage in project preparation (before the Consultants were involved) discussions were held under UNDP auspices to decide on the locations of the field trials. Agreement was reached in principle for the participation of India, Mali, Philippines and Sudan but in the event India did not participate in the field trials which were, therefore, hosted by and carried out in Mali, Philippines and Sudan. 1.3 Context of Review This Review is submitted to the World Bank at the conclusion of the Phase I of a UNDP Project to test and demonstrate small-scale solar pumps. During this Phase, which commenced in July 1979 and ran until May 1981, the Consultants completed an initial State-of-Art Report (Ref. 1) in December 1979; this Review is intended to supersede the earlier Report and therefore repeats many sections from it. Subsequently, after gaining the approval of the World Bank for their recommendations, the Consultants purchased a selection of the more credible photovoltaic and thermal small-scale solar-pumping systems available in early 1980 for field testing in Mali, Philippines and Sudan, in collaboration with the energy research agencies in those countries. Four systems were tested in each country; of these eleven were photovoltaic (PV) powered and one was thermal. Samples of the motors and pumps used in the field-tested photovoltaic systems were subjected to a parallel programme of laboratory testing in the UK, principaily to determine their performance characteristics. Sample PV modules from their arrays were performance tested in the UK and in the USA by independent testing authorities and the modules were subsequently subjected to intensive environmental testing in the USA. 3 Finally, the data produced from these test programmes were utilised in computerbased mathematical models of PV and thermal small-scale pumping systems. The purpose of the modelling exercise was to allow the rapid evaluation of the relative merits of the many different system options that are possible in making up a solar pump, as part of a design study aimed at identifying the most promising technical approaches to pump design for low head irrigation appUcations. Since the ultimate criterion for “goodness” of a solar pump is the actual cost of the useful pumped output it produces over its lifetime, an attempt was made to introduce the relative costs of different system components and subsystems into the modelling process and a parameter to assesscost-effectiveness was adopted. This Revievi necessarily draws on the general conclusions reached in Phase I of the Project and for the convenience of readem the Executive Summary of the Project Report is included in the Review as Appendix 3. It is hoped that the Report and Review together will provide guidance to governments and agenciesabout the performance and cost-effectiveness of many solar pumps and the desirable features which they should possess. 1.4 Scope of Review This Review disccssestte general technical and power requirements for smallscale irrigation and reports on a study of system economics, reviewing the effects of variation in the major influences on solar pumping and the differential movement in prices. The principal photovoltaic and thermal prime-mover power options available for small-scale irrigation pumping and the engineering requirements for this type of system are described. A technical assessmentis made of the main types of solar pumping system avaiiable and information is given on possible future trends. The Review concludes with a comparison of the relative efficiencies and costs of the main system options which are currently available and with an estimate of the costs for which it may be possible to produce these systems in the future. 4 2 ROLE OF SMALL-SCALE SOLAR PUMPING FOR IRRIGATION 2.1 The Incretig Importance of Irrigation As the end of this century is approached, feeding the worlds population (which is expected to increase by SO%by the year 2000) will prove an increasingly difficult and challenging problem, and nowhere more so than in the poorer, less fertile or densely populated regions of the developing countries. Irrigation is widely perceived as one of the key components in improving food production; there is no readily identifiable yield-increasing technology other than the improved seed-water-fertilizer approach. Further, it is expected that in the next two decades about three quarters of all the increasesin the output of basic staples will have to come from yield increases, even though during the past decade yield increaseshave only supplied half the increase in output, (Ref.3). This is largely becausethere is less and less uncultivated, but fertile, land available in the more densely populated regions; hence irrigation will be important not only to increase the yield from existing cultivated land, but also to permit the cultivation of what are today marginal or unusable areas of land. Table 1 indicates the irrigated areas in the world (Ref.4), together with the principal developing countries where irrigation is practised. The majority of the land brought under irrigation since 1972 is mainly in countries where irrigation has traditionally been practised. It is apparent from thin Table that only a limited number of countries have significant areas of land under irrigation, and that two of the largest countries in Asia, China and India, account for half the world’s irrigated land. It is also interesting to note that nearly half the irrigated land in the whole African continent lies in the Nile delta region of Egypt. These relatively densely populated, intensively cultivated regions almost certainly practise today what many other regions, with increasing population pressureson the land, will have to practise in the future, if sufficient food production is to be assured. Fanning in Asia today is perhaps the most realistic model of the kind of farming that will need to be prac”?*d in Africa and Latin America tomorrow. 2.2 The “Energy Crisis” and Irrigation A large proportion of the world’s traditionally irrigated land is commanded by gravity-fed water obtained by controlling the flow of rivers and providing suitable canal distribution systems. This is evidently a desirable method since there are little or no energy costs associated with distributing the water once the scheme has been completed. However, there are limits to the amount of land that can readily be commanded by gravity-fed water, and many populations and the land they require to cultivate cannot benefit from such schemes, now or in the future. -5 46of total REfRON &principal inipti0” countries 1.sou-m&soumRAsrAsIA China lndia P&iStaIl Indonesia TpiWaa Thailand PkiSppins Korea (others) 66 2. NORTR AMRRICA 9 3. NROPE 7 4. MIDDLE RAST 1w Iran Turkey Akhanistm (OthCIS) 4 3 2 ,I: 10 5. U.S.S.R 6. AFRlCA Em” Sudan M&WY Algeria ) ) Libya S. Africa ) (others) 5 5 3 1 1 1 (combined) 3 111-? 7. CARIRBEAN & CENTRAL AMERICA MC&X Cuba (others) 4 0.5 2 =h 8. SOW-l3 AMERICA [email protected] Chik PCN (others) 1.2 1.3 0.9 2 @h 9. AUSTRALASlA World Total 1.4 I 201.9 100 TAME 1 - IRRlGATED AREAS OF TRE WORLD IN 1972 6 During the 1950s and 6Os,the cost of petroleum-based energy tended to fall in real terms, which encouraged the increasing use of engine-driven irrigation pumps (and also the spread of rural electrification, and hence electrically energised irrigation pumps). In fact the area of irrigated land in the world has been estimated (Ref.4) as having increased by about 70% in the period from 1952 to 1972, and much of this will have been through engine or electric-motor driven pumping. Although mechanised irrigation continued during the 1970s since 1973 the great increasesin real terms of petroleum costs (and hence electricity costs) have reduced tbs mar&in to be gained by farmers from irrigation, because food prices cannot and have no: been allowed to increase in line with energy costs. The high cost of oil, in addition to putting an increasing burden on those farmers who already practise mechanised irrigation, severely discourages farmers fro.n bringing new land under irrigation. A further problem ln recent years has been the availability of diesel fuel. Supplies of fuel in many countries and particularly in remoter and poorer regions have become increasingly unreliable even for those with the resources to purchase them. This is a further inhibiting factor which worries many farmers who often depend on a supply of fuel to prevent the total loss of a crop. Some governments attempt to mitigate the situation by subsidising oil and rural electricity for use in agriculture, but many of these governments are the very ones that can least afford such a policy which incum balance of payments deficits, largely because of the increasing cost of oil imports. There is a pressing need therefore in many developing countries to discourage the increased use of oil even though there is an equally pressing need to increase food production, and mechanised irrigation is a widely recognised means to do this. Thus, it is becoming most important to find new methods for energising irrigation pumps that are independent of oil or centralised rural electricity. The latter is almost always generated from oil and the poor power factors, high infrastructural costs and peaky demands make it an unpromising source of energy for large, poor, rural communities. 2.3 SmalI-Scale irrigation Small-scale irrigation (sometimes known as micro-irrigation) is certain to become an increasingly important and widely used agricultural technique during the next few decades, particularly in developing countries. This is because the majority of land holdings in the poorer, more densely populated parts of the world are small, often a hectare or less. Studies have shown that these numerous and small land holdings are in fact more productive than larger farming units. An Indian farm management study (Ref. S), indicated that small family-run land-holdings are consistently more productive than larger farm units in yield per hectare, although the small units are more demanding in terms of labour inputs. A survey conducted in Brazil (Ref. 5) also illustrated the better iand utilisation of small farms, but family-sized land holdings only achieved this through applying 5 to 22 times as much labour per hectare as large farms. 7 SmalI land holdings also generally achieve better energy ratios (i.e. energy value hi the food product/energy inputs to produce it) in their crop production than is achieved by large-scale mechanised agriculture. This is discussed in detail in Ref.6 which indicated that typical energy ratios for tropical subsistence and semisubsistence agriculture arc in the range 10 to 60, while mechanised large-scale commercial agriculture generally has energy ratios of from 4 to less then unity. Hence, in a world ‘,, .th diminishing availability of commercial fuels, there appears more scope for signiticantly increasing food production through encouraging labour-intensive small-scale units which are likely to produce more fmm a given land and energy investment. Investment in inigation also has a positive effect in terms of alleviating poverty; for example, irrigation schemes can double the amount of labour required per hectare of land (Ref.5) and raise the incomes of landless labourem, even though the farmers of course derive the greatest benefit. The same reference gave details of surveys of average percentage increases in household income for farmers prac?ising irrigation, compared with those who do not. The increases obtained were 469% in Cameroon, 95% in South Korea, 90% in Malaysia, and 98% in Uttar Pradesh State, India. !n the Malaysian case, the benefit, on average, accruing from irrigation to landless labourers was 127%. The Asian Development Bank is assisting in the improvement of 12,000 sq.km of land through irrigation schemes, which are expected to benefit 5.1 million people and create 400,000 man years of work. When the potential of these current projects is reamed, they are expected to craate an annual increase of 3.2 million tonnes of unmilled rice: the per capita income for families which benefit directly is expected to increase from 92 to 227 US dollars, nearly 150% (Ref. 7). The scope for gaming improved yields through small-scaIe irrigation is substantial; for example the average rice yield in South and South East Asia is about 2t/ha, while in Japan, with sophisticated small-scale irrigation and land-management, yields are typically 6t/ha (Ref.7). The Asian Development Bank has reported that a doubting of rice production per hectare should be possible in the region within 15 years (Ref.7) and is considering a programme with this target. 304,000 sq. km of rainfed and 175,000 sq. km of inadequately irrigated~ land is to be converted to adequately irrigated land through an investment of 540,000 million US dollars (1975 prices), which is an average investment of about 11,000 US dollars per improved hectare (Ref.7). 2.4 Power Requirements for Irrigation The minimum power requirement, (P), to lift irrigation water is directly proportional to the product of the total state head (H), (or height through which the water must be lifted from its source in order to flow onto the fields, plus pipe friction energy and other energy losses), and the flow rate of the water at any given moment (Q). i.e. P = k.H.Q (where k is a constant) Generally, for economy, the system will be designed so that the static head will comprise a large proportion of the total head, so that losses arc kept reasonably small. The cost of pumping water is closely related to the rate of power usage (i.e. the energy requirement in a given time period). Hence the higher the static head or the larger the quantity of water to be pumped, the greater the resulting costs. a The static head is a Rutction of the vertical distance from the surface of the water source to the surface of the field. However, higher heads and consequently more power for a given flow rate are normally needed in order to distribute the water from the pumping system delivery pipe to the plants. The simplest method is for the water to be distributed by gravity, which means that it has to be pumped to an elevation high enough for it to flow effectively across the land area to be irrigated. The outlet for a gravity distributed irrigation system may need to be typically Im above ground level; while this is low in absolute terms, for low lift irrigation from a source say only 3m below ground level, this extra metre of head requires 33% more energy (and similar associated costs) to move a given quantity of water. Pumping water through distribution pipes also implies adding to the effective head, due to friction in the pipes. The use of pipes of too small a diameter can have a profound effect on the energy requirements (and consequent costs) of low head pumping systems. Sprinkler irrigation and most drip feed irrigation systems, although economising effectively in the water quantities needed, generally require quite high operating pressures, and are therefore expensive in terms of energy requirement at low static heads. So the effective head through which water must be pumped depends on the total static head, the lossesin the pipework linking the pump to the source and delivery point, and on the additional head necessary for the water distribution system. The quantity of water needed to irrigate a given land area depends on a number of factors, the most important being: 1. 2. 3. 4. 5. 6. 7. a. nature of crop crop growth cycle periods climatic conditions type and condition of soil land topography field application efficiency conveyence efficiency water quality Many of these vary with the seasonsand the quantity of water required is far from constant. The design of a small irrigation pump installation will need to take all these factors into account and include consideration of the economics of providing storage. The estimation of overall irrigation water requirements starts with the water requirement of the crop itself. Calculation begins with a standarised criterion known as the ‘reference crop evapotranspiration” (ET,:: this is the rate of evapotranspiration from an extended surface of actively growing tall green grass completely shading the ground and not short of water. It depends on temperature, humidity, wind and cloud cover (or sunshine hours). ET, is the water demand of the reference crop itself and if water is available from no other source represents the quality which has to be supplied by irrigation. ET, can be computed by a number of internationally accepted methods using standard meteorological data (e.g. Ref 38) or estimated from pan evaporation data. Since ET, depends on climatic factors it varies from month to month, sometimes by a factor of 2. 9 The evapotranspiration of a particular crop (ETcrop) is of course different from that of the reference crop and is determined from the equation. ETmop = ET, x K, Kc is a “crop coefficient” which varies with the type of crop, stage of growth, growing season and prevailing weather. It commonly varies from about 0.3 during initial growth to as much as 1.Oduring the mid-season growth peiiod. Thus the actual crop water requirement ETcrop can (& does) vary quite considerably during the growing season. The nett irrigation requirement over a specified time period is the depth of water required to meet the crop evapotranspiration demand, less any contributions from rainfall, groundwater or stored soil water. (Some rainfall is lost to the crop by surface runoff, deep percolation and evaporation and the rainfall is factored to obtain the ‘effective rainfall’ available for crop use). Allowance also has to be made for the water needed for preparation of the land - this can be significant, particularly in the case of rice. The nett irrigation requirement then has to be increased to alIow for losseswhich occur during conveyance and application in the field to obtain the gross irrigation requirement. Table 2, adapted from Ref. 8, illustrates typical irrigation water requirements for cotton and wheat in the vicinity of Lake Chad in central West Africa. The estimated availability of solar energy, on average, for the various months is also given. The Table illustrates a typical irrigation pattern to be expected in a semiarid tropical region and shows clearly how the actual irrigation demand varies substantially with the growing seasons.The Table also shows that if solar energy is to be used to power a pump, the sunniest months do not necessarily coincide with the months of maximum demand. The same information is presented graphically in Figure 1, A constant field application efficiency of 60% is assumed in Table 2: this is the proportion .of water applied to the field which actually contributes to the nett irrigation requirement and is a function of the method of water distribution and of the farmer’s water management abilities. Surface (flood) irrigation is typically 30 to 60% efiicient, while sprinkler irrigation systems can be 60 to 80% efticient. Certain new systems of irrigation, such as trickle or drip irrigation or under-soil irrigation can be better than 80% efficient. Typic& figures for other crops and regions are in the range from about 4,000 cu. m./ha per crop (using an efficient distribution system and good water management) to as much as 13,000 cu.m./ha per crop in the Sahel dry season,i.e. 400 to 1300mm of water per crop. (Refs. 2, 8, 9, etc.). Typical growing cycles are of the order of 120 days under tropical conditions; the averagedaily requirement is thus in the range 35 to 100 cu.m./ha. Assuming an averageof 8 hours pumping per day is possible (as would apply with solar pumps without any energy storage), then the average flow required is in the range 1.2 to as much as 3.5 litres/sec per hectare. For solar pumps, the flow under peak sunlight conditions (of around lOOOW/sq.m) is likely to be about 25% above the average,so peak flows in the range 1.5 to 4.3 litre/sec are likely to be needed for solar-powered irrigation systems to service one hectare of land. NOV DEC JAN FEB MAR APR MAY JUN JUL AUG SEP OCT TOTAL Referencecrop evapottanspitation(ET,,) mm 179 146 144 142 208 228 227 185 181 148 151 199 2138 Averagerainfall mm 0 0 0 0 0 0 8 9 69 142 43 9 280 Effectiverainfall (ER) mm 0 0 0 0 0 0 0 0 35 71 22 0 128 Gmundwster co&ibution (GE) mm 0 0 0 0 0 0 0 0 73 31 0 0 104 WHEAT-)) Cmpping pattern N- Cmp mef&ent (Kc) 0.6 0.8 1.0 0.7 COTTON (( - )) 0.6 0.6 1.0 1.0 0.9 - - - Crop Waterrequirements ETerop=EToxKc mm 107 117 144 99 0 137 136 185 181 133 0 0 1239 Net irrigation requirement @%rop -ER-G,) mm 107 117 144 99 0 137 136 185 73 31 0 0 1029 Gms ligation requirement (Nett a I .67) mm 179 19s 240 166 0 228 221 308 122 52 0 0 1717 Cms itrigaticn tequitement mm/ha 1790 1950 2440 1660 0 2280 2270 3080 1220 520 0 0 17,170 Meanflow rate&ctam (lb./s) 2.1 2.2 2.7 2.0 0 2.6 2.5 0.6 0 0 - Volume required/hectare (asumittgpumping for8h/day) m3/d 60 63 77 59 0 76 73 0 0 - Averagesolar energyper day kWh/m2 6.1 5.6 5.6 6.4 7.0 7.3 6.9 5.8 6.4 - 3.6 1.4 103 39 6.4 6.2 17 5.5 TABLE 2 - 1RRlGATfONWATER DEMANDAND SOLAR ENERGY AVAILABILITY - LAKE CHAD REGION FOR COTTON AND WHEATCROPPINGPATTERN (After Reference8) IFallow LEGEND ~ ~;;~&f;‘“;,:“,‘bgn:h. Gross irrigation requirement (proportional to anarpy demand and drown q 01 to indicate rcbtion between energy demand and cner~y available) FIGURE 1 - SOLAR ENJZRGYAVAILABILITY AND CROP IRRIGATION WATE” .~taM&qD (LAKE CHAD REGION) 12 Assuming, for example, a water-table Sm below the pump discharge level and distribution thereafter by gravity, then the peak hydraulic power output corresponding to the range of flow rates given above will be 73 to 210 W/ha. Slightly different numerical assumptions can yield hydraulic powers of up to 3OClW/ha,for a Sm lift. Input power requirements will be considerably higher depending on the system efficiency, as will be discussed in more detail later. Power requirements are directly proportional to the actual head across the pump. This consists largely of the static head through which the water is lifted, but also includes the pipe friction and associated energy losses. The friction head can be a significant proportion of the total head, particularly with low head systems of 5m ‘or less or with systems having a significant length of delivery pipe. The velocity head is normally small unless water distribution is by spraying. Hence it is important to take the head losses in the pipework into account as part of the system design. Since water for irrigation has a ftite value, related principally to the marginal value of the extra crops gained, and the cost of water increases with increases in head, it becomes decreasingly profitable to irrigate as the pumping head increases. There will be a certain static head above which it will be uneconomic to irrigate, as the costs will exceed the benefits. Similarly, transmitting the water long disrances will increase the power requirements because of the extra energy required to overcome pipe friction*. If pipe friction is avoided by distributing the water through channels by gravity, extra head has to be provided to obtain a sufficient hydraulic gradient in the channel. Also, in the latter case,evaporation loues will increase in relation to the surface area of water in the channels and unless the channels are totally impervious (i.e. concrete, steel or plastic), losses due to seepageare also possible, all incurring a power demand. It follows from this that the losses (either by leakage and evaparation, or by an effectively increased pumping head) will increase in relation to the area of land to be commanded by the pumping system, and any economies of scale obtained by using a larger pumping system will be countered by extra costs of distribution. From the foregoing description, it will be appreciated that careful system design is necessary if solar pumps are to be used effectively in irrigation applications. 2.5 Pumping Methods Available 2.5.1 General The principal traditional methods for lifting irrigation water itI the tropics have been (and in many areas still are): 1. 2. human labour, (e.g. using a Shaduf water crane) draught animals, (e.g. “sing a Persian Wheel or Noria) * The head loss in the pipework due to friction is a function of the resistance coefficient, the size of the flow rate, the iength of pipe and the inverse of the fifth power of the pipe diameter. The resistance coefficent is related to Reynolds number and the relative roughness of the pipe surface. 1. Human Labour Readily available. Low investment costs. Can be flexibly deployed. High feeding costs and associatedcosts (wages). Very low output. Divens a valuable resoutce from more productive xtivities. 2. Draught animals Readily available. Medium investment costs. Convenient power output for stnaU.scale irrigation. Can be flexibly deployed. High feeding costs involving extra food production. Feed required even at times when no power can usefully be employed. 3. Petrol ot die& CueSedsmall engblcs Widely available technology. High outputs possible on demand. Pottable. Low initial capital investment cost per unit of output. Fuel costs dominate and are increasing in real terms. Fuel shortage mumon in many areas.Spare parts often hard to obtain. Good maintenance difficult. Relatively short useful life. Breakdown common. High imported element involving scarce foreign currency in most countries. 4. Centmlised rural ekctrification Low marginal bl”es,nw”t cost for prime mo”er (electric pump) if transmission Linesinstalled. Pump sets are reliable provided supply is guaranteed. Can be low co%>depending on power source and system. Electricity supply often unreliable in Third Wodd due to pesky demand, low load factors, spaxe consumer population Very higb system investment coat and high generating and distribution costs. Extended power failures could cause widespread crop loss. 5. Wind pumps Relatively matute renewable energy technology, when used for stock watering. Low cost in areaswith adequate wind regimes. Zero fuel cos,s. Low maintenance needs. Not yet well developed for irrigation purposes. Moderate output fluctuates with wind. High investment costs. 6. Water wheels, turbines. tam pmlp and current turbines Low cost, IongJife, low maintenance, fuel-free power source if suitable site conditions ate available to exploit water power. Depends on tslatively rare site conditions which limit the ateas that could benefit froin this type of prime nKw*. Potentially tow cost renewable energy techno. ~y&pecially if agricultural waste can be used Fuel requites land use or transport of coal. Steam engine technology generally obsolete so modern equipment not readily available. Safety problems with boilen. Constant or frequent attendance needed. 8. Biogas fuelled small engines AUows advantagesof small engines given above but independent of supply of petroleum fuel. Immature technology for biogas production, but has ~owcost potential. Fertilissr produced as a by-product from digester. 9. Solar radiation (converted directly) Energy resoutce almost universally available in areaswhere irrigation is useful. High correlation belween energy availability and needs. Zero fuel co% Lortg life and low maintenance are possible. r Includes most disadvantagesof small engines (spates and maintenance). High watet requirements for digester. Feedstock for digester may be scarce, panicularly in arid regions. Fairly high labour needs for digester. Quite higb investment costs in digester and gas storage. bnmature technology with high investment cost at present. Diffuse energy resource compared with most alternatives will ensure fairly higb investment cost at best. even after development, Output subject to solar insolation variations. TABLE 3 -COMPARISON OF PRINCIPAL METHODS OF IRRIGATION PUMPING 14 Since the 1960’s, relatively prosperous farmers in many areas have made increasing use of the following forms of mechanised pumping: 3. 4. diesel, gasoline or kerosene fuelled internal combustion engines mains electricity energised electric pumps Water power has been used traditionally in hillier regions to command land not accessible through contoured gravity fed canals; the main techniques using this resource are: 5. 6. 1. water wheel powered Persian Wheels hydraulic ram pumps water turbine driven centrifugal pumps (mainly in China) Other technologies not generally used as yet, but which may have future potential, are: 8. 9. 10. 11. 12. biomass (or coal/lignite/peat) fuelled steam engines b&mass (biogas/alcohol/plant oil) fuelled ix. engines windpumps (widely used at present for livestock but not much for irrigation) water current turbines (under development) solar radiation, directly converted (which is the subject of this Review) The main advantages and disadvantages of these methods are outlined briefly in Table 3. 2.5.2 Solar Pumping Options There have been a number of books and papers published recently which include reviews of the present state of the technology of solar-powered pumping (see for example References 16 to 24 inclusive). The information given in these publications has been supplemented and updated by the extensive enquiries to manufacturers and research institutions made under this Project with a view to purchasing equipment for testing under Phase 1. The most feasible options for solar-powered pumping systems for the power range under consideration, with the present state of development of solar power technology, are illustrated schematically in Figure 2. At present, none of these alternative systems has a clear overall advantage although sane options appear to be better than others. Their various advantages and disadvantages are discussed in detail in later chapters. Examples of all of these systems (with almost ewry permutation of engine and transmission illustrated) are cr;i+ently being developed but few are commercially available; the majority of those on the market consist of Fixed photovoltaic flatplate arrays powering a d.c. electric motor/pump unit. The systems and components which are commercially available today are reported on separately in the Project Report. In addition to the main options for solar pumping referred to above, a number of further technical options which may possibly have a future role are reviewed in Chapter 6. COLLECTQRS ENGINES -1 1 Electric TRANSMISSIONS Actuator PUMP FIGURE 2 - FEASIBLE OPTIONS FOR SOLAR - POWERED PUMPING SYSTEMS 16 2.6 The Suitability of Solar Pumps for Irrigation The main advantage of using direct solar radiation as a power source for irrigation pumping is that it is the only energy resource that is almost universally available both when and where it is needed. Most of the countries where irrigation is necessary have high levels of solar irradiation and the availa5lity of sunlight does COP relate at least partialiy with the crop water demand. The correlation is often not perfect, in the sensethat the time of the year when most solar energy is available often does not exactly coincide with periods of maximum irrigation water demand, (this is illustrated by the example in Table 2 and by Figure 1 derived from it). However, the period of least availability of solar energy is normally coincident with the period of least or zero irrigation water demand, being invariably the rainy season. A solar pump, once installed, is clearly independent of fuel supplies or other external constraints and, in effect, runs on the sameenergy resource a%the plants it servesto water. However, most of the lifetime cost of a solar pump has to be met at the time of purchase, since the main cost element is inherent in the first cost, operating and maintenance costs being very low for a successful solar pumping system design. The provision of finance to assist the poorer farmers to meet this fit cost is an important aspect of the transfer of this technology that will have to be addressedonce it is clear that solar pumps provide genuinely economic means of raising water. Government backed agricultwal credit schemes have been in operation in a number of developing countries and it is not thought that this represents an insuperable problem. The economics are discussed in Chapter 3. In addition, the high cost of present day solar pumping systems can be expected to fall substantially through improvements in performance and design, and through economies resulting from an increasing scale in manufacture. On the national level, the widespread deployment of solar pumps would help to improve food production (and alleviate rural poverty in the process) largely through the better use of local resources of land, labour, ground water and sunlight. This would be achieved without the increasesin petroleum imports (with consequent’ balance of payments probiems), which would be needed if ’ conventional oil fuelled engines or rural electrification were used. The maximum benefits on the national level would be obtained if a high level of local added-value can be introduced into the manufacture of solar pumping systems to minim& the foreign currency requirements associated with the introduction and subsequent widespread use.of the technology, as discussed in section 3.4. Further possible advantagesof solar pumps are that they are potentially long lasting and reliable, if the technology is developed to maturity, compared with small engines, and the maintenance, servicing and repair skills should be less demanding and more easily taught then those needed for engines. Hence they are particularly well suited for use in remote areaswhere communications and accessare difficult so that a system that is relatively independent of external supply lies has a significant advantage. 17 Also, for smaU-scaleirrigation pumping duties, most land-holdings are sufficiently small to demand no more than a few hundred watts of pumped output power at low heads. Small engines do not scale down well to this power level, being inherently inefficient and lacking in durability when rated at much less than I or 2kW; also the maintenance costs of small engines are a major factor and there are no significant economies to be found in these when comparing say a 2kW engine with a 5kW one. Therefore, for very tight pumping duties it is hard to use a small engine effectively; usually a larger than necessary machine must be used, while solar pumps on the other hand do scale down without serious diseconomies. Hence solar pumps are likely to be more competitive with internal combustion engines forsmalbscale applications and this is confirmed by analysis of available data. For example Ref. 8 fmds that a 1 ft3/s solar pump would be more competitive against a similar output engine than a 2 ft”/s unit (= 56.6 lit&.) which is substantially larger than the systems under investigation for this Project. Some of the main disadvantages inherent in solar pumps, are: 2.7 0 that they represent unfamiliar technology. 0 they are still, in the main, immature and therefore of high unit cost (due to the present limited scale of production). 0 that standards of reliability, efficiency and longevity are in general not yet at the level that is required for their widespread adoption. 0 although efficiency will improve with development, and costs will come down, solar-powered systems will inevitably remain large in relation to their power rating due to the low power density of sunshine and the relatively low efficiency with which it can be converted to shaft power for pumping. Thus they are likely to involve more materials and take up significantly more land area than engines powered by fuels, or even than windpumps in areaswith sufficient wind to make the latter viable. Size and Efticiency Considerations for Solar Pumps 2.7.1 Size.Considerations For a constant flowrate, irrigation becomes more uneconomic as the head increases due to the corresponding increase in power demand and hence unit cost of water. The ternn of reference of this Project assumed that the likely head for irrigation pumping would be in the range 5 to 10m as there is much land available having a water table as close as this to the surface. The majority of farms in developing countries are less than two hectares in area, and a large proportion are under I ha. As discussed in section 3.2. this is precisely the size range where engine powered pumps are generally oversized and too expensive, and hence this is the best opportunity for solar pumps to prove competitive (see Figure 9). Individual solar pumping systems, and particularly photovoltaic (PV) systems, do not exhibit marked economies of scale, but economies may be found in the volume of production. Hence by manufacturing larger numbers of small units, costs are likely to come down faster, and if this happens it is possible that the unit costs for small PV systems will be lower than for larger ones simply due to the scale of manufacture. 18 Further benefits from the developmenr of small solar pumps would also be found because: larger land holdings can make use of several such units (while still benefitting from the !:w costs resulting from mass production for a Largemarke:). average distances water has 3 be transmitted would be lower, and consequently the averap: total pumped head, and water lossesthrough seepageor evaporation would be reduced. Hence the power requirements and the unit cost of water would be reduced. smaller solar pumping units could be introduced progressively, thereby reducing the capital sums to be raised to several smaller amounts rather than one large sum. a number of smaller units will have a higher reliability, since a failure would not be so serious as when only one pump was in use. the task of water management is substantially easier with a number of small units, so overall water usageefficiency could possibly be improved. Small land holdings of less than 0.5ha are expensive to irrigate, but solar pumps offer a better prospect than engine pumps. This is shown in Figure I I and discussed in detail in Chapter 3. If solar pumps are to be successful, they will require reasonably efticient water distribution and management, which implies a peak water demand probably in the 40 to 80 m”/day range per hectare. In practice, and especially while the technology remains expensive, it will be important to hold the water requirement at the bottom of this range if irrigation is to be economically viable. Most water management techniques or improved distribution systems are likely to be cost-effective if they significantly improve water application efficiency and hold down the water requirement. One point in favour of small land holdings is that field losses are likely to be smaller (as a percentage of water pumped) and water management is likely to be easier. Figure 3 illustrates the relationshiop between head, flow rate and pumped power output and indicates the requirements of a typical 1.Oha land holding. 2.1.2 Water utilisation efficiency A solar pump has to be sized to he able to satisfy the peak irrigation demand for a given area and these conditions may only last for a relatively short time: thus at other times of the year the pump is likely to have excesscapacity. Figure 1 illustrates this using the example from Table 2 of energy availability and irrigation water demarid, for two crops per year S- 7- ‘ROPRIATE 5. 1 \ \ I \ HOLDING 432I- I 0 I FRXIRE 3 - POWER OUTPUT REQUIREMENTS FOR VARIOUS HEADS AND DEUVERY RATES 20 (wheat and cotton) in Chad. It should be noted that the area chosen as an example is likely to be more favourable than most for the use of solar pumps due to its low rainfall and high irradiation levels. It is noticeable in this example that if a solar pump is sized so asjust to satisfy the demand for irrigation water in the month of maximum demand (June in this example), then considerable surplus capacity will be avallable in most other months. This is due in part to: 0 variation in crop water demand during the growing cycle. 0 the changeover between crops (when no irrigation water is required). 0 ram supplementing or replacing irrigation water The n&t result of this mismatch between power availability and demand is that (m this example) only about 40% of the solar energy available can usefully be applied for irrigation. This example is likely to be reasonably typical; the situation would be better in terms of percentage water utilisation when there is a shorter time between changeover of crops or when there is lessrain to supplement the irrigation water. The situation could be even worse since the systems will have to be oversized even for the month of maximum demand to allow for: 0 years with less then averagesolar energy availability. 0 the fanner being unable to be present to make effective use of all available solar energy in the month of maximum demand. 0 solar pumping systems being unable to use the proportion of solar energy received with a power level lower than that needed for pumping to commence. The proportion of available solar energy that can be used will be dependent upon the climate, crop cycles chosen, type of solar pump and effectiveness of the water management practised by the farmer, but it will never be possible to use all the water that can be pumped with the available energy for an application such as irrigation. Factors that wiU increase the proportion of solar energy that can be used include: 0 the use of storage to improve easeof water management and reduce the peak flow requirements of the pump. 0 multiple CrOFFitlg 0 use of water for other purposes in addition to irrigation. 0 operation in arid regions with even less rainfall then in Chad. 21 A further pOsSibility would be to fmd an alternative use for any surplus electrical or shaft power. As the cost of solar pumps falls, the degree of oversizing that can be tolerated will become greater, but initially it seemsthat if they are to be successful it will be in areas where serious oversizing due to extreme variation in irrigation water demand can be avoided. Although it was beyond the scope of PhaseI of this Project to consider in detail the utilisation of irrigation water after it leavesthe pumping system, it must be emphasised that the detailed development and subsequent introduction of small-scalesolar pumps in realistic pilot projects cannot be conducted without consideration of this vital issue. The way the water is used can have a profound effect on the size of solar pump required for a particular duty, and hence on the cost; there are indications that an efficient distribution system coupled with careful usagecould very likely halve the size and cost of pumping system required. Both time and money should therefore be invested in field water storage and distribution systemsaimed at improving water application and utilisaticn efficiencies, as well as in providing the necessarytraining for at least the first groups of farmers to use this technology. Such investments are likely to earn a good return in terms of reducing the sixes of solar pumping systems needed to perform a given duty as weU as ensuring that the practices learnt by those farmers who pioneer the use of the technology are the most appropriate. 2.7.3 System efficiency The complete “system” for the purposes of analysis includes: 0 the solar resource (incoming solar radiation) 0 the solar pumping system hardware 0 the distribution system “hardware” and water management “software” ” The contribution of each to overall system efficienty is discussed below a) The solar resource As explained in section 2.7.2 it is not possible to utifise more than a proportion of the solar energy resource that is available. There are two principal methods for reducing the proportion of solar energy received which is so diffuse as to be below the system starting threshold; these are to reduce the threshold or to extend the period in the day when the irradiation received by the collector is above the threshold. 5 -TIME FIGURE 4A - VARIATION OFIRRADIANCE OPTIMUM MCLINATION LEVEL RECEIVED BY A FIXRD COLLECTOR AT TIME [CURE 48 - VARLKITON OF IXRADLQICE LEVEL RRC!ZIVED BY A SUN-TRACKING COLLECTOR 23 To reduce the starting threshold involves making the system more efficient and avoiding any hysteresis in the system, such as might be caused by static friction. For this reason, low static heads centrifugal pumps are attractive as their starting torque can be lessthan their running torque. Although attempts have been made to maximise the use of available solar irradiation, most manufacturers (whether of small photovoltaic or of small thermal solar SYstemS) have favoured a fared array for simplicity of manufacture and operation. Hoccver, Figure 4A shows that this arrangement produces a variable irradiance level through the day closely approximating a cosine curve. This cute is such that a considerable period of sunshine in the early morning or late afternoon can be lost so far as a solar pump is concerned, becausethe attenuation due to misalignment of the collector with the sun’s rays reduces the available power below the starting threshold. For example, when the array is more than 60 degreesfrom normal to the sun, sunlight is attenuated in power density by SO%,(cos 600 = 0.5). Figure 4A also shows how if the collector is not normal to the sun’s beam radiation at midday, there is a further attenuation. Usually such a collector is fixed facing south in the northern hemisphere or facing north in the southern hemisphere with its plane inclined at an angle to the ground equal to the latitude of the location. This places the plane of the collector parallel to the axis of the Earth so that it receives solar radiation exactly normal to its surface at noon on the equinoxes which is the fixed position which maximises the solar energy received over a whole year (in theory, neglecting seasonalweather effects). This presents problems near the equator, where the collector ideally should be set horizontally, since it should also be set at a sufficient angle for rain water run-off to wash away dust and dirt from the collector surface. At noon in mid-winter and mid-summer, the array will be approximately 23 degreesaway from normal, to the sun’s ray, which is the equivalent of about 1 h 30 min before or after noon at the equinoxes: this represents an additional attenuation factor of around 8%. The problem of attenuation of received irradiation can be largely corrected through the use of an appropriate design which permits the collector angle to be adjusted seasonally; it is fortunate that cosines of small angles are quite close to unity; IO degreeslack of normality represents only 2% attenuation and it takes a full 31 degreesof divergence from the normal to attenuate the received irradiance by even 10%. A substantial increase of usable solar energy can be gained by arranging for a flat plate solar collector to be movable so as to track the sun. If the array is tracked from east to west, following the sun so that the array is always normal to the sun’s rays, then in theory the full intensity of sun tight can be experienced (of around IOOOW/m’) from the moment the sun is above the east horizon until it starts to go beiow the western horizon at dusk, (which is the broken rectangular outline indicated in Figure 4B). In IGURESA - VA~UAT~~NOF~RRADL~NCELEVELRECEIVEDBYCOLLECTORREPOSITIONED ONCEPERDAY tWorific4ing ~~.EsB times, - VARIATION~FIRRADIANCELEVELREC~DBYCOLLECTORREPOSITIONED 'IWICEPERDAY 2.5 practice, even in very clear desert climates, the considerably increased thickness of atmosphere traversed by sunlight in the periods immediately after dawn and before sun-set serve to attenuate the level of radiation received, even normal to the sun’s rays. As a result the solid cul~e in Figure 48 indicates the typical variation in irradiance level through the day if measured normal to the sun. Obviously climatic conditions can greatly modify these curves, but given clear conditions, they are close to reality. By comparing Figures 4A & B, it can be seenthat a perfectly tracking collecter will achieve a given threshold irradiance level earlier in the da% and permit a longer running time for a solar pump compared with one having a ftxed collecter. Also, extra energy can be coilected due to the greater power density available in the sunlight normal to a collector. In simple geometric terms the extra energy gained by perfect tracking can be 50% or more. The gain can be greater in certain areaswhere it is commonly more cloudy in the middle of the day. In addition certain solar systems can achieve a higher average conversion efficiency if the input energy variation, between the starting threshold and the maximum, is kept small. Perfect solar tracking is difficult to achieve, as it requires that the colletter be driven with precision through 180 degreesin 12 hours, and that it varies in inclination as well as in azimuth due to the seasonalmotion of the sun’s apparent path across the sky. It is possible to maintain close to the optimum orientation of a collector by tracking it in a single plane normal to the earth’s axis, (i.e. by rotating it about an axis incline zo the horizontal at an angle equal to the local latitude). Occasional seasonal adjustment increasing the inclination by up to 23 degreesin mid-winter and reducing it by the same in mid-summer can marginally improve the collection of irradiation. As wiU be explained later, really accurate tracking is only necessary where concentrating collectors with high concentration ratios are used, in order to keep the focussed image of the sun on the absorber area. The’provision of automatic and continuous tracking can greatly complicate what otherwise would be an extremely simple, fixed system, and introduce both extra costs and a lower level of reliability. Therefore, it is interesting to consider a useful compromise involving the occasional reorientation of a movable “fixed” collector, by hand. Figures 5A and 5B show the effect of moving such a collector, once per day and twice per day, respectively. In both casesit is assumedthat the collector is optimally orientated and that it is moved at the correct times. it is clear that moving such a coUecter once per day extends the running time for a solar pumping system having a given starting threshold to almost the sameperiod as with continuous tracking: early morning and late afternoon performance gains would also be expected. However, a certain proportion of the peak midday &radiance that would be picked up by a fixed array is lost. Figure 5B shows that moving a couector twice per day even allows the midday peak irradiation to be collected and results in about 95% of the irradiation available with continuous tracking to be received. Clear’-! there is substantial potential in tracking, either continuously or occasionally. At least 40% extra energy can become available through the use of some form of coffecter reorientation or tracking. b) Solar pumping “hardware” efficiency The main objective is to achieve the most cost-effective rather than the most efficient system. Efficiency, however, is crucially important in order to minimise system costs, providing it is not pursued so far that diminishing returns set in. Certainly, the more efficient systems field tested under this Project also 3 to be the most cost-effective. Figures 6 and 7 ilhtstrate the instantaneous power flow through a typical PV and a typical thermal small pumping system. With PV systems most of the lossesoccur at the cells in conversion of light to electricity, while in thermal systems most of the lossesare shared between collector heat lossesand the rejected heat and other losses from the heat engine. In either case,only about 4% of the input energy appearsin the form of hydraulic output. c) Distribution lossesin irrigation As previously explained, the distribution system wiU involve losses, either due to water that fails to reach the roots of the crop (through , leakage, percolation into the soil or evaporation) or due to the energy used in providing the extra head necessaryto distribute the water. With very low static heads, this latter loss can be substantial. Bad management can result in further losses, through water not being directed correctly and running to waste, or thrc ~gbthe system not being used at times when it could be, thereby losing solar energy that could have been usefully applied. 2.8 Alternative Applications for Solar Pumps There are two major problems in applying solar pumps for irrigation applicatiom0 the low cost for which water must be delivered (less than 6 US cents (1981) per cubic metre). 0 the relatively variable demand for water,often with high but brief peak demands and long periods with no demand at all. Both of these require a lower cost system than can readily be manufacturered at present and either subsidies or large and speculative investments are needed to produce the breakthrough to high volume manufacture and low unit costs that is needed. I I 0 Pipework lossas 0.5% pl IOU heeds ) unit) FIGURE 6 - LOSSES IN A TYPICAL SOLAR PHOTOVOLTAIC - PUMPING SYSTEM F / / ”:::., .:i !gii, iji: i FIGURE 7 - LOSSES IN A TYPICAL SOLAR THERMAL PUMPING SYS’IEM 29 However, the provision of pumped water supplies from boreholes or wells in remote and arid or semi-arid regions wig tolerate rather higher investment costs (and unit water costs? than irrigation and also tends to bs an end-use with a steadier year-round water demand. It is likely that solar pumps, even at cwrent prices, might be at or close to economic viability for such applications, providing the pumping head is not too great. It is precisely for this kind of application that the autonomy of a solar pump is a valuable asset;the low level of maintenance that is (at least in principle) possible coupled with long life and no fuel requirements makes the technology po:entiaUy attractive to those instihrtions responsible for maintaining remote water supplies. It was beyond the scope of PhaseI of the Project to look at applications other than irrigation, but these will be considered in the future phasesof the Project. It may be that through the initial use of solar pumps for premium applications like water supply, the scale of manufacture will develop to a level where prices drop and the prospects for using solar pumps for irrigation substantially improve. However, most water supply applications will be through high pumping heads (greater than IOm) with the consequent deployment of positive displacement pumps, whereascentrifugal pumps are more appropriate for the low pumping heads (up to IOm) which are generally required for economic irrigation pumping. Therefore, although there is considerable overlap in the technical requirements for both end-uses,there are also some significant differences. Another important possible application for small-scale solar pumps is to provide water for cattle and other livestock in semi-arid regions from boreholes. Again, a higher unit water output cost is tolerable and the demand is steadier than for irrigation. A possible virtue of small-scalesolar pumps for this application, in addition to the obvious ones of needing little in the way of external supplies or maintenance, is that they can be sized economically to produce a much smaller output than is readily possible with engines. A major problem with the provision of water for livestock in developing countries is overgrazing that can result from excessive concentration of animals around large-output watering points. The provision of a greater nuinber of smaller watering points would mean that herds could be limited in number and hence more efficient use made of the available grazing. Pumping with a small flow rate over a longish time period, as with a solar pump may also have advantagesin areaswhere there is a large draw-down on the well at higher flow rates, or where the well may even be pumped drywhen using an engine (which can causeseveredamageto the well and pump). It is worth noting that to achieve the declared goals of the current IIN International Drinking Water Supply and Sanitation Decade, new water supplies will have to be commissioned to provide for the needs of an average of half-amillion people every day of the decade until 199I. There seemslittle doubt that solar pumps could and will play a useful part if a programme as ambitions as this is to succeed. 30 3. ECONOMIC AND TECHNICAL FEASIBILJTY 3.1 I”iWd”CC0” The previous chapter has reviewed the potential role of small solar pumps in irrigation practice and deliberately avoided more detailed discussion of the economic, technical and developmental questions which need to be resolved if solar pumps are to be used successfully in practice. This chapter is in three parts: 0 system economics n technical requirements 0 importance of local manufacture In the section on system economics a description is given of an economic evaluation of and comparison between solar pumps and engine pumps designed to irrigate 0.5 ha. A baseline model was fust developed and run, and a sensitivity analysis was then performed to test the way in which the economic picture was affected by variation of the more important assumptions implicit in the model. The section on technical requirements is based on the experience of the Consultants on the Project with the systemsundergoing field tests. It illustrates that attention to practical design points, simplicity of operation and reliability is as important as high efficiency. The final section in the chapter discussesthe critical role of local manufacture of part assembly in the introduction of a new technology into a developing country and the difficulties of implementing such a policy. 3.2 System Economics 3.2.1 An economic model for sensitivity analysis It is difficult to make absolute economic judgements on small-scale solar pumps because not only is the technology immature but also evaluation is made difficult by the variability and uncertainty of many parameters that affect the pump system economics. One of the economic guidelines proposed by the World Bank for the Project was that the cost of irrigation water delivered to the field should not be more than 80.05 per cubic metre (1919; the equivalent fgure for 1981 may be taken as $0.06 per cubic metre). This was taken to be a globally representative fgure which, if attained, would mean that the extra income obtained from the additional crops would yield a reasonable return. The adoption of this figure does not imply either that in certain areaspumps providing water at a greater *costwill not be viable or that, if the cost is lower the pump will necessarily be economic. 31 Despite such uncertainties it is possible to set up a plausible “baseline” economic model and use this as a tool to investigate the sensitivity of solar pumping system costs to variation of different parameters. Such a model can also be used to indicate the relative costs of a solar pump compared with alternative and competitive options. The analysis was conducted in purely economic terms and in principle considered all the costs to the economy regardlessof who incurs them. Financial costs, e.g. subsidies and taxes, were excluded because these represent a transfer of money within the economy rather than a cost to it. In many cases there may be no truly economic means to produce increased food crops, in the sensethat for social reasons food needs to be underpriced, particularly in countries with poor populations. So the fact that in many cases neither engines nor solar prumps appear to be economicaUy viable needs to be viewed in that context. The economic analysis was based on 1981 US dollars prices for present day costs, allowing for inflation. at a rate appropriate to each of the individual inputs to the projected cash flows, rather than differential inflation.* A real discount rate of 10% was assumed when compared to the general rate of inflation. While the absolute cost figures in the model will mly be indicative, they can be used to draw helpful comparisons on the :osts of delivered water, as wall as on their relativities, and their sensitivir I to alterations in key state variables or design parameters. This model ias been used to compare the effects of varying different key parameters on the relative costs of small-scale solar pumps and small engine driver: pumps. The reason for comparing solar pumps witb engine driven pun-as is that much of the current interest in solar-powered pumping stenl~ from the substantial increase in petroleum-based fuel costs during the last decade; hence such a comparison gives an indication of the soundness (or otherwise) of substituting solar pumps for engine-driven pumps, but in itself will not demonstrate the economic smmdness of either alternative. For the purposes of examining the general parameters that affect solar-pumping system costs, the model considers one of the most costeffective PV systems, representative of best current practice, and applies suitable assumptions as to the likely future inflation rate that will affect replacement systems and maintenance costs. Any radical technical developments that could affect the economics through step changes in typical performance or costs are not considered; such possibilities are discussed in following chapters and would of course make solar pumping that much more viable. 3.2.2 “Baseline” Model Assumptions The analysis for the model was run on a computer which was programmed with the parameters indicated in Table 4. These parameters are solar pump related or engine related or external parameters which are independent of I. * It is appreciated that it is more usual to make economic analyses at constant (1981) prices, allowing only fcr differential movement in prices, but in this case which involved inflating different cash streams at different rates it was more convenient to do the calculations in the way described. 32 BASE-LINE MODEL: 0.5 ha SOLAR PUMP PARAMETERS System Unit Cost Period for Economic Analysis of Solar PUmP Annual Maintenance Costs Year 1 Design Operating Life Av. Subsystem Efficiency Overall System Efficiency ,S lO/WPeak = Computed System Life in Years fi 100 15000 hr 40% 4.4% ENGINE PARAMETERS Engine Rated Power Derating Factor System Unit Capital Cost Period for Economic Analysis of Engine PumP Annual Maintenance Costs Year 1 Estimated Operating Life Engine Efficiency Subsystem Efficiency Overall System Efficiency 3 kW Peak 0.667 S 6OOikW = Computed System Life in Years S 0.50 /h run 2000 h or up to 7 years maximum 15% 40% 6% BXTERNALPARAMETERS Discount Rate Ave. General Inflation Rate (including engines) Ave. Fuel Price Inflation Rate Ave. Solar Pump Inflation Rate Ave. Water Delivery Head Water Del. Head at Pk. Demand Time Ave. Daily Water Demand for 0.5 ha. Peak Daily Water Demand Ave. Daily Irradiation Ave. Irradiation at Peak Water Demand Fuel Cost at Year 1. 10% 15% 5% 5m Sm 20 m3/day 50 m3/day 6.1 kWh/m2 6.4 kWh/m2 S 0.4/litre COMPUTED RESULTS Actual Annual Energy Demand Required Array Peak Rating Solar System Capital Cost Solar Pump Utilisation Factor Solar Pump System Life Engine Capital Cost Engine Fuel Consumption Engine Running Time Per Year Engine Operational Life Cost of Water in Year I : Solar Pump Cost of Water in Year I : Engine Pump 99 kWh 319w s 3193 0.4 1I Years $ 1800 165 Litres/Y ear 124 Hours 7 Years I 1.46 c/m3 8.58 c/m3 TABLE 4 - PARAMETERS FOR BASELINE MODEL 33 the type of pumping system. A 1981 s,stem unit capital cost of US dollars S IO/W peak was assumed, where peak watts correspond to the array rated electrical output in an irradiance of 1000W/mZ with a PV system. Prices of the most costeffective systems currently available are in fact approximately twice to three times this level but it would not be realistic to use prices relating to systems that are manufactured in the small individually hand-assembled quantities currently applicable for solar pumps, and then to compare the apparent economic costs with such a mature product as a mass-produced diesef pumping system. The price of S IO/W pk is used becausethat appears to be the economic price of a small-scalesolar pump (of around 200-300W array output in peak sunlight conditions) if it is manufactured and shipped in quantity. The breakdown is approximately S7IWpk for the army and S3/Wpk for the Balance of System (BOS). The latter consists of a motor at typically S300, a pump at S250, and it is assumedthat the packaging, electrical connections and pipework plus the array support structure could be provided for no more than around S 200, giving a total of S 750 or S3/Wpk with a 25OWsystem. It is also assumedthat various optimisations .discussedin the companion Project Report, which can improve the system cost-effectivenesswill have been successfully implemented to achieve the perfonnancelcost requirements necessary for S 10/W pk. There is obviously some uncertainty about this assumption, but the value of a sensitivity analysis is that if the baseline assumptions are thought to be wrong, the effect of changing them can readily be assessed(seeexample of Figures9, lOand II). In the analysis, the system life is calculated from the assumedoperating Life and the calculated number of hours per year of running time required to meet the water demand. In the baseline model an operating life of 15000 hours is assumed;there is no guarantee any present-day system will actually achieve this, but a life of this order is necessaryif reasonably low costs are to be achieved. Motor brushes (if fitted), motor and pump ‘bearings, and almost certainly the pump impeller will have to be replaced during this period, but an averageannual maintenance cost of S 100 is assumedto allow for this. Also the array, which is a major cost element, will quite probably exceed the life-time requirement assumed. . Finally, the solar pump subsystem (motor and pump) is assumedto have an overall efficiency of 40% (on average)which is in line with tl¶e best small subsystems currently available. It is assumed that the baseline system efficiency is at the highest level currently achieved by the best small-scalesolar pumping systems(4.4%), since any production model will require to be well optimised if it is to succeed. The period used in the economic analysis to calculate the system annualised capital cost was taken as equal to the system life. The baseline engine is assumedto be the smallest diesel type generally in widespread use, i.e. of about 3kW rate power. However, it is derated to 67% of peak power (this is the averagerunning power assumed,including idling time) and the averageengine shaft output efficiency is taken as IS%. Various references were consulted for costs and the consensus arrived at was about S6OO/kW(rated). Useful references for this and other parameters were (81, (9), (IO), (ll), (12) and (13). As with the solar pump, the fife of the engine is calculated from its estimated operating life (typically 2000 hours for a small diesel in tropical field conditions). Maintenance costs are taken as SO.50 per hour, which is a reasonable averagefigure especially as they do not greatly influence total annual cash flows (see the sensitivity analysis curves of Figure 9). The value of 40% assumedfor the engine subsystem efficiency includes the pump, pipework losses(at low heads the pipe function and related lossescan be a large proportion of the total head for engine pumps) plus any mechanical transmission if the engine is not directly coupled to the pump. This is not a best, but an averageefficiency, and is in line with the practical results such as reported in Reference 15 with regard to kerosene-fuelled irrigation pumps in Sri Lanka. The external parameters consist of economic assumptions on discount and inflation rates, together with information on the water demand, head and the two energy resources - irradiation per day and fuel costs. It is worth emphasising that the assumptions on discount rates and inflation are not generally applicable but are simply taken as reasonableeconomic values to test the sensitivity of the costs of the two systems to variations in them. The overall period taken for the baseline analysis should be a whole multiple of the Livesof each of the capital items which require replacement: in this case 11 years for the solar pump and 7 years for the engine. The baseline analysis was done for a period of 22 years, allowing for one solar pump replacement and two engine.replacements. The eight year life of the second replacement engine wiU have negligible effect on the figures. The discount rate should be real (i.e. in a model using inflated cash flows greater than the inflation rate by the discount rate chosen) and reflect the opportunity cost of capital. Therefore, for the baseline model, the differential between the discount and general future inflation rate is taken as 10%. Fuel price inflation, which has risen substantially more than that (of the order of 25% per annum in the last few years) is taken as 15%. Becausesolar pumps are an immature technology, with falling costs in real terms anticipated as the production volume of solar PV cells in particular increases, the inflation rate applicable to them wiU be lower than the general rate of inflation, and substantially lower than the likely inflation rate for petroleum-based fuel costs. The inflation rate applicable to the replacement costs of solar pumping systems has been assumedto be 5% i.e. at half the general infiation rate. This is an unavoidable weaknessin the analysis as it certainly would not be correct to assume that solar pump costs will increase in lime with general inflation, but there is no certain way of judging whether the averagerate chosen for the baseline model is realistic, although it is believed that it is not seriously incorrect. 35 The pump operating head for the baseliie model is taken as 5m as a representative vahm. At lower heads the application will be more economic, but the costs of irrigation as higher headsmay not make it a widely applicable technology for some time. The land area to be commanded by the baseline model was taken asOSha, as this requires a solar-pumping system capable of being portable, of a not unreasonable capital cost and which would compete particularly well with engine powered pumps as they are least competitive for such small, but common land holdings. The market for a system of this size would appear to be sufficiently large to permit significant economies to be made through massproduction. The model’s irrigation water demand and its variability are for a mean water requirement (over the year) of 20m3/day (40m3 /ha. day) and a maximum of 50m3/day (100 m3/h&day). This compares with 47 and 103m3/ha.day for the Chad example previously used by way of ihustration (seeTable 2). The Chad data are of interest as they represent an area which should be favourable for the use of solar powered pumps due to its low rainfall and high irradiance levels. The baseline data are representative of a slightly less, but still favourable area. As indicated in Figure I, only a fraction of output can usefully be applied to the crops; this is the Utilisation Factor, (given in the “computed results” of Table 4). The size of system needed to meet the specified maximum monthly water demand is calculated for the irradiation conditions specified for that month. From this, the Utilisation Factor is calculated by dividing the averagewater demand specified for the year by that for the month of maximum demand. This factor gives a measureof the oversizing oi a solar pumping system required to cater for the variability of irrigation water demand. With the baseline example the Utilisation Factor is 0.4 (i.e. the averagewater demand over the year is 40% of that in the peak month); in the Chad example, the Utilisation factor works out at 46% and there are probably few places where a more favourable factor than this might be found. The baseline engine fuel costs were set at S 0.40/l&, which is a reasonable economic cost for 1981 but which makes no allowance for inland transport (to deliver fuel in the field) and storage costs that will possibly add substantially to this figure in remote areas. Shortages of diesel fuel in many developing countries cannot be addressedby this simple economic model. However Figures 8 and 9 allow the effect of higher fuel costs to be taken into account and are not large for such a lightly used engine as in the baseline model. These costs have no bearing on solar pump economics per SC Subsequent runs varying the parameters to allow an assessmentto be made of what would happen with a different baseline from that chosen “are described later. 36 737 73, 73, 737 737 737 1262’ 1262 ,262 L262 126: 1262 1262 ,762 It62 1262 1262 37 3.2.3 Baseline Model Results Various key properties of the resulting system were calculated and these appear as “computed results” at the foot of Table 4. The annual cash flows of the solar and engine pumps for every year from 198 1 to 2002 comprising the sum of the uniform equivalent annual costs of capital, annual maintenance costs and annual fuel costs, are shown in Table 5. The World Bank’s figure of 5 c/m3 (1979) for water has been inflated at the general rate for comparison with an averageunit cost of water calculated from the annual cash flow for each year on the b&s of a daily volume of 20m3. It must be appreciated that the annual cash flows (and the derived figures for average unit cost of water) depend on maintenance of the 22 year perspective for the economic analysis; if a 5 year or 10 year period were taken the results would be different. Steps in the equivalent annual coat of capital occur at years when replacement systems are required and the capital sums for replacements have been increased at the appropriate inflation rate. The model calculates the Present Worth (PW) for each of the cash flows for the two options over the 22 year period of t’le analysis. The PW assessmentenables an economic comparison to be made of the two sets of cash flows, each of which contains different patterns of expenditure over time. The results are given at the foot of Table 5. For the baseline assumptions, over the 22 year period, the solar pump option has a PWjust greater than that for the engine. In the view of all the uncertainties inherent in such analysis this slight difference is of no significance and on the basis of these assumptions there is nothing to choose between the systems on economic grounds. The averageunit cost of water each year is indicative of the way in which the economics of the two systems are moving relative to the target cost of water (inflated at the general level of 10%). As will be seenthe trend is for the solar pump economics to improve while the economics of the engine pump deteriorates. If the analysis had been conducted over a shorter period the PW of the engine pump annual costs would have been lower than those for the solar pump, and if it had been taken over a longer period (ignoring the great uncertainties in so doing) the PW of the solar pumps would have had a more marked advantage over the engine. 3.2.4 Sensitivity Analysis Key parameters in the baseline model were systematically varied to investigate the sensitivity of the comparison between solar-powered and engine-driven pumps to such variations. A programme of 28 such variations was completed, as indicated in Table 6. In all such computer runs, the key factors outlined in the column headed “description” in Table 6 were changed, although in some casesseveral parameten had to be changed to make the new model realistic. ,7 37 37 37 17 ,7 37 37 37 37 n 3, 37 ,I 37 3, n ,7 ,I ,7 ,7 37 ,7 37 37 17 37 37 TABLE 6 - RESULTS OF SENSITIVITY ANALYSIS 40 22 34 22 *4 18 43 30 34 20 40 26 44 24 40 53 80 39 Some of the changes investigated in the sensitivity analysis m wt TVat ‘he system lives of the solar pump and engine pump (consistent u’l:h c Levi ting hours of 15000 and 2000 - seeTable 4) would not be I 1 years and 7 years, as in the baseline analysis. To simplify the sensitivity analysis a uniform period of 20 years was adopted: even though this may not correspond with whole multiples of the lives of the two pumps, the error will be very small and make no difference to the conclusions. Some key results of these runs are summarised along with the descriptions in Table 6. 3.25 Discussion of the Indications from the Economic Model Anal. sis a) General The absolute values obtained from the model can only be regarded as indicative, due to the large number of assumptions made, but the trends indicated by varying individual parameters are generally fundamental to the economics of small solar and small engine-powered pumping systems and are of potential significance in evaluating such systems under different conditions. Table 6 provides a useful summary of key indicators that vary with the changed parameters. For example, factors that change the actual energy demand (i.e. changes in head or gross water demand) cause significant changes in solar pump capital cost and in engine fuel costs. This is a fundamental difference between the two systems; the solar pump only has a finite daily energy input per unit area of collector, so changes in demand require changes in size and hence capital cost, but the engine can run more or less to suit demand and simply uses more or less fuel in proportion*. A second-order effect for the engine is that maintenance costs and the periods between replacement engines are a function of the average number of operating hours per day, which varies with demand. Large variations in demand will dictate the need for a larger or smaller engine, as is discussed in more detail in the analysis of the results that follow. The effects of varying key parameters are perceivable from Figure 8, which shows the annual operating costs of selected systems. Figure 8 also illustrates the effects of different assumptions on inflation and discount rates on the unit water costs. Some of the more significant influences on the unit cost of water, revealed by the analysis, for solar pumps are set out in the form of a sensitivity diagram in Figure 9 and these are discussed in turn in the sub-sections below. * This is the case for the comparison made in the analysis because the minimum practicable size of the engine (3kW) was still much more powerful than needed for the duty proposed. If it had been technically feasible to adopt an engine size which required it to run for 24 hours, the economic analysis would have demonstrated that installed power and capital cost were material factors. IO0 90- w- GASELINE ASSUMPTIONS ( ABRIOGEOI Genaral. Discard rote 20%. Inflation rate 10%. Static head 5m. Mter demand ZOm!/dd/day(averageI. Fuel costs year I 4Ocm?s/lilre IIS% inflation) lrmdiation 6.1 kWh/duy (averape). s. capiml cd 1981 $10~ O&p, life 15,OOJhr. System efficiency 44%. peak. m- E~Q~O. copilolcost1961 t600/kw Oniqn life 2,OOOhr. System efficiency 6.0%. 60 SO- ‘IO- 30- I i! ,M’.. YEAR. ;URJZ 8 - EFFECTS OF INFLATION AND DISCOUNT RATE ON ANNUAL CASH FLOWS [email protected] lsxprewd in ferns of average unit cost of water) 26 ASELINE MODEL ASSUMPTIONSj ,ncml. gly& vcynr rm.2 20% Capitol cost #600/ kW, fltiion rote 10% Mointenonce cosf SOcentshr. Oeriqn life 2,OOOhr. Sq5tem efficiency 6% Rated power 3kW. ,bX Iptt.1 cost $IO/WPK. oilltcnom cost $1OO/yr. [email protected] life 15,OOOhrs. wtem efficiency 44%. lted pwter 32OW. FIGURE9 EFFECTS OF CHANGES I -I. 2 -2. run. 3- 3. 4-4. 5 -5. 6-B 7- 7 G-G. 9-9. IN i Head. Waler damond(land or.?a commanded) Solar pump utilishm factor (mlio Peak lo “7~” YO,.?~ demand). Discount rot.2 Solar pvmp system efficiency. Solar pump SyJtm co51 per pedl watt (Of orray outpu, ), Operotinq life. Mointenonce costs. Engine fuel costs. -RESULTSOFSENSlTlWTYANALYSISONSOLARPUMPS 42 Variations in Pumping Head The solar system capital costs and the engine system fuel costs are directly proportional to head, all other things being equal. Engine maintenance and capital costs are sensitive to head variation, as the higher the head the longer it has to run per year and it therefore requires more maintenance and more frequent replacement. By contrast, solar maintenance costs are a fixed sum and independent of head. However, the solar pumping system has to be resized if the head is changed, but the engine need not be (it can slmpb~run longer per day), so the capital cost ratio of solar pumps to engines increaseswith head. In other works, a “baseline” solar pump at 2.5m head is less tb?:, the 1981 capital cost of the baseline engine, but one fi~r 10m has to be nearly four times the baseline engine cost. Variation in the head also effects the Present Worth ratio, (i.e. life-time operating costs ratio), and the unit water costs of solar compared with diesel, year by year. Increasing the head, according to the model, makes solar increasingly uncompetitive with the engine mainly becausethe engine has to be worked harder, so its “load factor” improves. Figure 10 shows how variations in head affect the unit water costs and how these relate to the assumedunit capital cost of the solar pumping system. It shows that for a solar pumping system to achieve the required 6 c per cum. at 5m pumping head, a unit capital cost of only $5/W pk is required if all the other baseline system assumptions apply. Presently available systems costing over $20/W pk can only be economically viable (on the 6clm3 assumption) for headsof little more than lm. Variations in Gross Water Demand This has a similar effect on the relative capital costs of the baseline systems as varying the head, since it also causesa pro-rata change in energy requirement. i.e. the water demand for 1 ha requires four times the size of solar pump and four times the fuel for an engine pump as for 0.25 ha. Hence the capital cost ratio between solar and engines diverges with increasing land area or water demand. The affects of water demand (or area of land commanded) on unit water costs are shown in Figure 11, for various assumptions on solar system unit capital costs and for the baseline and the baseline doubled engine fuel costs. 28 26 la 16 PUMPING HEAD (d FIGURE lO-EFFECTOFPUMPINGHEADOFWATERUNITCOSTS 46 reeson to consider the use of solar pumps for small land holdings rather than large ones, (see Table 6). The capital cost ratio virtually increases pm-rata with land area, from about unity at under half a hectare to 3.55 at 1 ha (and 7 for 2 ha, and so on). d) Variations in Utilisation Factor The sensitivity diagram, Figure 9, indicates how increasing the solar pump Utilisation Factor by smoothing demand variations wiU significantly reduce unit output costs. A doubling of the Utilisation Factor reduces water costs by 17% (either by requiring a smeller system or by making a given system more productive), while halving it will increase unit costs by 70% It is of great importance with solar pumps to achieve a good Utilisation Facto] through applications with as consistent as possible year round water demand. Engines on the other hand do not cost so much when they are no: in use, and so poor Utilisation Factors do not significantly affect their water costs as much as solar. Therefore, solar water pumps become decreasingly competitive with engines in situations where the demand radically varies through the year e.g. where there is an extensive rainy season. e) The Effect of Different Inflation and Discount Rates Figures 8 and 9 show that the annual cost of solar pumps are highly sensitive to discount rate assumptions. Figure 8 indicates that halving the baseline discount rate gives a 30% reduction in annual costs, while doubting it gives a 70% increase. Halving the discount rate from 20% to 10% effectively means a nil return on capital invested since the general inflation rate is 10%. Although this is unlikely to occur it is valid for the purposes of sensitivity analysis and shows how the economics of solar pumps become more favourable compared with engines if the opportunity cost of capital is low. As noted in section 3.1, it was not within the scope of the economic analysis to consider the financial strategy required to make the pumps affordable in the context of the introduction of new technology to a developing country. Such a financial strategy would consider reducing the tinanciai cost to the farmer by subsidised capital costs or interest rates and would represent a transfer of resources within the economy: this would only be considered once the basic economic viability of the pumps had been established. Figure 8 also indicates the substantial effect different fuel price inflation rates have on engine annual operating costs: an increase from the baseline assumption of 15% fuel price inflation to 25% causes engine annual operating costs virtually to double by the 41 year 2000, by which time the engine operating costs would be over four times the baseline solar pump costs, and three times the acceptable level for h-rigation pumping. The effect of assuming different fuel costs at year 1 has less significance for engine pumping system output costs in year 1 than might be expected due to the very small demands on the engine which result in fuel being only a small proportion of costs. In year 1, with the baseline example, fuel is oniy 13% of total engine running costs, although by 2000 it is over 30% even with only a 5% differential between fuel and general inflation. With kerosene or gasoline engines, which are of lower capital cost but lower efficiency, fuel would account for a considerably greater percentage of total costs. I 0 Solar Pump System Operational Life Assumptions Reducing the working life of the solar pumping system has a major infhtence on costs. This is shown in Figure 8 where a system with an assumedlife of “only” 7500 hours running time (cf. the baseline at 15000) will have been replaced twice as often and amortized over half the period, resulting in annual operating costs approximately twice those of the baseline system. 7500 hours is actually quite a respectable lifer for most machinery and 15000 hours is exceptionally long for small machines, (for example, cars typically have an operating life of the order of 2500 to SO00hours). Therefore a lot of attention to detail will be required from solar pumping system designers if such long lifetimes are to be reliably achieved. Figure 9 illustrates what will happen if such long lifetimes are not achieved: for example a reduction to “only” 6000 hours r?s&s in a 70% increase in unit water output costs: the sameas increasing the baseline solar pumping system unit capital costs from $10 to 6 17. In practice, it seemslikely that the array could last much longer than the motor and pump unit, (possibly with a life of 20 years or more). This could be offset against subsystemswith a shorter life than that indicated desirable by the present analysis. Attempting to increase the operational life from 15000 hours, which would be technically difficult, appearsto give diminishing returns. Solar pumping irrigation systems should be designed and constructed for a life in excessof 10000 hours if they are to have the best prospects of becoming economically viable. 48 8) Solar Pump System Efficiency Varying the subsystem efficiency has the same effect as varying the solar pumping system overall efficiency by the same proportion. Any technical improvements which affect system efficiency will have a similar economic effect, whether they involve just the subsystem (motor and pump) or whether they include the PV may. For one model run an improvement in average subsystem efficiency from 40 to 50% was trisd which yielded a 17% improvement in costs. This is just technically feasible through good design, but must be close to the upper limit for such small motors and pumps. More profound are the negative effects of worse than the baseline system efficiency. A 20% reduction in efficiency, from 40% to 32% (typical of many current solar pumping systems) results in a 26% increase in unit costs. Some of the poorest subsystems tested under the UNDP Project field test programme were only around 25% efficient; these would have unit output costs over 90$/bgreater than for the baseline model. Lower efficiencies than this would be quite catastrophic to the economics. h) Solar Pumping Systems Considered as a “Mature Product” Ii the unit capital cost of solar pumps falls substantially due to mass production, then the opportunities for further price reductions will become less marked and solar pumps will begin to increase in cost more in line with the general rate of inflation; in other words they will have become a “mature product”. In one run a higher inflation mte of 10% was envisaged for the solar pump replacement costs to emulate the effect. This is what would tapper! if sc!ar pcrrzps ceased tc, get any cheaper in real terms than E IO/W pk ir 196i. The result is a I 10% increase in soiar pump Present Worth which makes the solar pump just fail to deliver water within the 6c/mJ limit for most of the period considered. A more likely scenario if solar pumps ever develop to maturity is that their costs might drop to around $ S/W pk (in 1981 real terms) and then cease to get lower in real terms compared with the general inflation rate. Under this condition (run 20 in Table 6). a favourable Presest Worth ratio is also obtained against the engine pump and evw the solar pump tint-cost is lower than that of the engine. 3.2.6 Conclusions from the Baseline Model Analysis It appears that solar pumps will need to satisfy a number of economic and technical requirements if they are to pump water at an economic unit cost. Key requirements are:- 49 a) solar pumping system capital cost down to be under SSjW pk (in I98 1 real terms) when product achieves maturity (full volume production). b) solar pumping system life to average in excess of 10000 hours (preferably about 15000 hours) before any major replacement costs are incurred. C) solar pump subsystem efficiency to be at least 40% and preferably better (giving an overall efficiency of at least 4.5%) over a reasonable range of operating conditions. Until solar pump capital costs fall in real terms to less than about $5/W pk, solar pumps will generally only be economically preferable to diesel pumps for small land-holdings having average water demands in the 10 to 30 m3 per day range. It is possible to combine features identified by the model as helping to reduce the costs of solar pumps to achieve earliest utihsation. For example, improved system efficiency, combined with increased life and offered at a subsidised price or with a soft loan in the interests of producing cheap food; ii in addition the technology is initially popularised in very low head applications in small land-holdings where the water demand is relatively steady, then further cost reductions may be expected. 3.2.7 The Capital Cost Barrier to the Introduction of Solar Pumps From the point of view of the economy of the country as a whole, the criterion for choice of solar powered pump or engine pump should be the PW of the annual cash flows over the twenty year period. These are given in Table 6 for each of the sensitivity analyses: any PW ratio (solar/enghte) of less than unity means that a solar-powered pump represents the more economic option. This condition occurred for higher fuel costs, lower pumping heads and water demands, lower discount rates, more efficient solar subsystems, increased solar pump lives and lower solar pump capital costs. However the capital costs of solar-powered pumps are a crucial factor in their adoption by farmers, since the capital cost is the predominant influence on the average unit cost of water delivered. For this reason the capital costs of solar pumps and engines were compared for each of the sensitivity analyses in the last three columns of Table 6. The fit cost of the solar pumps was lower than that of the engine in only three instances: when either the pumping head was halved (to 2.5”) the water demand -.Tl_,- _-r the capkil iost irss halved (equiva!ent to an L7iga&.. aF$a Cc ‘ nY*““.*, of the solar pump was halved to $5 per peak watt; the first two of these analyses were equivalent of course to the halving of pump capacity and first cost. SO This capital cost barrier is one that can only be overcome by action taken by the Government or related national institutions. Farmers of small land-ho!dings in developing countries are very poor and possessno capital resources of their own, nor do they have ready accessto private institutional timance.As was noted in section 2.6 once Governments are convinced that solar pumps represent a genuinely economic means of pumping water, consideration can then be given to providing the finance necessaryfor the purchase of the pumps on terms which the farmers can acceptThis will probably mean a subsidy of one kind or another, implying a transfer of financial resources within the economy of the country: the acceptability and extent of such subsidy will depend on the policies of the Government towards the agricultura! sector, the improved agricultural yields that may be obtained, the energy situation of the couniry and any reduction in oil imports which may result. Government backed agricultural credit schemeshave been in operation for a number of years and their extension to the provision of solar pumps would not involve any change in principle. However, it is not yet certain that the various technical and economic barriers to solar pump usage for irrigation will necessarily be overcome. Solar pumps are probably already economic for certain more remote water supply applications (as distinct from irrigation applications) where higher unit water costs are acceptable; it may be that the necessaryboost to their development through quantity production will come initially through their wider use for water supplies in developing countries rather than immediately for crop irrigation. 3.3 Technical Requirementa 3.3.1 General The detailed aspectsof solar pumping system design are covered more fully in the companion Project Report and in the sections that follow, but certain general points are worth making at this stage. Any pumping system for general use by farmers in developing countries (or for that matter by farmers anywhere) must be capable of reliable operation with minimal maintenance under harsh operating conditions. All equipment must therefore be constructed to the standards required for a long and reliable life under agricultural field conditions and major system components should achieve a useful life of the order of 10,000 hours or more. To do this requires robustness and high quality of construction. In addition to aU the normal hazards of general wear and tear and corrosion faced by machinery, solar pumping systems(in particular PV arrays) are likely to be subjected to rough treatment and carelessoperation: also, there is no guarantee that the operator will respect manufacturers’ instructions, so the system should be effectively fail-safe; there should be no modes of self-destruction due to possible operatorerror. 51 ! It should be remembered that the points made in this section refer to equipment purchased in late 1979 and early 1980. Many manufacturers are of course aware of the need for improvement to their products and future progresswill be considerably aided by collaboration between them and the usersof their products, and the international agenciesand their consultants employed on projects of this type. 3.3.2 General Limitations of Current Equipment During the course of the Project the Consultants purchased and assessed the performance of a number of small-scalesolar pumping systemsunder tield conditions (in association with leading local institutions) in Mali, Philippines and Sudan. As expected this work indicated that solar pumps can be an attractive technology, with few mechanical components to go wrong and needing little maintenance. However severalproblems were encountered in installing and commissioning the systems,and some of these point to ways in which the technology might be improved. Also a few of the systems performed badly or unreliably due to either conceptual or detail design CliTOt5. It should also be added that both the Consultants and the various host country institutions involved provided professional engineers to supervise the assembly, commissioning and runrdng of the systems (the Consultants Resident Engineers also visited the relevant manufacturers’ factories to witness the systems being tested prior to shipping). Despite the well trained personnel involved and the close connection to the supplier, many systems presented difficulties in commissioning of one kind or another - most were minor, but in one or two casesit proved possible to get the systems to run only after modifying them in the field, and then with difficulty and with performance below specification. In view of these problems the Consultants believe that some improvements are necessary before such systems can be assembledand successfully operated by small farmers in developing countries, although the potential for achieving the necessarydegreeof reliability and simplicity is certainly there. AU but one of the systems tield tested were solar photovoltaic (PV), the only thermal system field tested was relatively immature and suffered a number of teething troubles that made the results of the tests inconclusive. This imbalance does not necessarily imply that PV systems are inherently superior to thermal systems, rather that it is more difficult to improvise a thermal system from off-the-shelf components because virtuaUy all components require to be purpose-designed.Thus most of the fmt-hand practical observations which can be made at this stagerefer to PV systems (although some would be common to either type). 52 3.3.3 Technical Concepts Some examples ioilow which illustrate that there are likely to be good and bad conceptual approaches to small-scalesolar pumping system design. For example, a large proportion of the systems tes,tedin the Project used su:face-mounted suction-pumps, possibly because the specification was for low head applications (in line with the needs oismall-scale irrigation), often pumping from surface water or from open wells. It appearsmore straightfonvard to design a suction pump system for such installations and also, it is easier to construct such a system by combining off-the-shelf components (this may have more to do with the manufacturer’s convenience than with the needs of the fmal user). However, a problem with suction pumps is priming them and maintaining their prime. This proved difficult and compromised the performance of at least two of the PV surface-mounted centrifugal suction pumps. Loss of prime can readily occur through a leaking footvalve. It only needs a particle of grit to lodge between the footvalve and its seat for all the water to drain back from the system as soon as the pump slows down or stops due to the sun being obscured briefly by a cloud, or at night. Some of the pumps that more readily held their prime and avoided associatedcavitation problems were vulnerable to damagein the event of running dry through loss of prime. One otherwise satisfactory pump burnt out its gland packing through running dry and thereafter leaked quite badly until a new seal was fitted, and another got so hot that it caused plastic pipe fittings to soften and distort. Only one manufacturer attempted to make their suction pump fail-safe by supplying a float switch to cut off the system if the water level fell too far but which would not cut off the system in the event of simple loss of prime. The only precaution another manufacturer took was to attach a lable to the pump to warn against running the pump dry. Although the positive-displacement surface-mounted suction-pumps that were supplied self-primed more effectively and tended to maintain their prime, they are inefficient at low heads and also relatively large and expensive. Only three of the PV systems tested incorporated submergedpumps but in the Consultant’s view these are the only options for this application. This was confirmed in practice, as the two best performers proved to be submerged pumps. One other rzment of system design which needs more attention from system suppliers, is the type of installation envisaged;is everything to be bolted down on concrete foundations or is it possible to make a portable system that can easily be carried into place and quickly installed. Only 53 one supplier, provided the latter type of system, which was more straightforward to assembleand commission as a result. All the others required the casting of concrete foundations which may be needed if a considerable length of piping full of water has to be supported or where the stability of arrays under high winds has to be considered. Bolted down equipment may be more secure from theft, but on the other hand a portable system can be taken into the fanner’s courtyard at night for safekeeping. Solar-powered systems are commonly specified by their power rating under peak sunlight conditions (usually given as lOOOW/mr) and a number of manufacturers compromised the performance of their systems under the more common it-radiance levels of about half peak sunlight conditions. The problems of system optimization are discussedin detail in the relevant chapters, but it would seem better if the trend of rating systems under peak sunlight conditions were changed to specifying the daily pumped output, perhaps under a standard solar day of say 6kWh/ma of received solar energy. In the end it is the volume of water that is pumped in a day, rather than the flow rate under peak sunlight conditions, that matters. A number of systemswere supplied with inadequate instructions, components missing and components assembledincorrectly by the manufacturer. This was partially due to the prototype status of some of the systems tested, and the relatively short delivery time allowed. The needs and problems of the manufacturer still appear to weigh heavier at this early stagein the development of the technology than those of the end-user. For systemsto be tailored more to the end-user will demand less reliance on “off-the-shelP’ construction and more emphasis on the “purpose designing” of components, a trend which may even make such systems initially more expensive due to the component development costs that will be incurred coupled with the small initial volume of production that is likely. In the meantime, the use of “off-the-shelf’ components will no doubt haVe an important role to play in to popularising the technology and in gaining further field experience, providing serious conceptual errors that could bring the technology into disrepute are avoided. 3.3.4 Detailed Design Points One problem requiring further attention is the vulnerability to damageof PV arrays and solar collectors for thermal systems, particularly in transit from the factory. The array is the most expensive single subassemblyfor a PV system, and it generally usesa glasscover. In the caseof thermal flat plate systems breakageneed involve no more than the cost of replacing the glass, but with PV systems a broken cover generally meansan 54 unusable module. Some of the modules used were rated at as much as IOOW peak, which implies a replacement cost of around S 1,000. A number of breakagesof this kind were experienced during the first Phase of the Project, both during delivery and in the field. Two simple solutions to this problem might be the wider use of high quality plastic module covers (such as polycarbonates), which would perhaps mean a marginal loss of performance, and a marginal increase in cost, but a large improvement in durability. Alternatively, instead of the toughened glass used by several PV manufacturers which crazes if impacted, laminated automobile glassnight prove superior (itis already used in one or two casesfor this purpose).Another improvement would be the use of smaller and more numerous modules; this would nave the disadvantage of involving more external electrical connections, but a single breakage would be less serious becausethe proportion of power loss would be small and the module would be cheaper to replace. Smaller units are cheaper to pack and transport and lesslikely to suffer damagein the first place. Such an arrangement would also allow more flexibility in the choice of array voltage-current combinations, giving a better chance of optimising the system performace more closely for different conditions. No concentrating solar collector systemswere tested during PhaseI (none was available to suit our specitication), but the sameobservations would of course apply to systems involving glassmirrors. There was a lack of concensusby manufacturers as to the correct inclination for optimum performance from the system. There are problems in the tropics at equatorial latitudes due to the shallow almost horizontal positioning that is optimum for solar energy collection, but bad for rainwater run off and self-cleaning. Here an adjustable array has obvious benefits, especially if it could be calibrated with recommended positions for different seasonsfor the region in question. In addition, some arrays were too close to the ground, making them prone to collect dust and ah more likely to be damagedby people or animals. Wiring presented numerous problems, with badly (or in some cases wrongly) prepared wiring harnesses,poor instructions and it was often difficult to connect terminal blocks to connecisn. No detailed analysis has yet been completed on wire losses,but it is believed these may be significant in many cases.Blocking diodes were another serious energy drain in some cases(and in two examples the systems tiere supplied with some or all diodes connected back to front). If follows that system arrays should ideally: 0 be adjustable in inclination and azimuth (or be fully portable) 0 have simple foundation requirements allowing for imprecise positioning with the array held well clear of the ground. Where loading conditions permit arrays should be fully portable and should not require any foundations. 0 have clear and simple instructions 55 0 have generous wiriig and effective mechanics! tem:inal grips to ensure good electrical contact. If possible blccking diodes should be avoided or be sized for low losses. One system was supplied with a length of 25mm diameter hosepipe at the delivery of the pump. If this had not been changed to SOmm,a significant loss of performance would have resulted due to pipe friction. Systems with suction centrifugal pumps can incorporate almost unclearable airlocks in the suction pipe-work due to poor design; one was partially cured by inverting the pipe and tilling the space where an air bubble had been. However, this is not the kind of modification that ought to be required. Another system was marred by a series of failures of electronic components in the electric motor commutation and in the power conditioner due either to overheating or to excessive transient voltages or currents. The system in question appeared sound in concept, and the circuits that gave trouble are worth including in the design-if they can be made reliable. The faults have been corrected but clearly electronic circuits incorporated in solar pumping systemsneed to use components with a tolerance of high ambient temperatures. They will also benefit from a degree of inbuilt protection from excessivetemperature, voltages or currents which should trip-out the system if it overheats. Operating voltages were generally around 6OV. Lower voltages are undesirable, as they require a low-voltage high-current array with consequent greater resistive line losses,or the need for heavier and more expensive cabling to compensate. However, d.c. voltages greater than about 80V present an electrocution hazard (dc is more dangerous than ac) and for safety reasonsare to be avoided; one system tested used 12OV dc, but it is understood the manufacturer is changing this to a lower voltage. Hence it appears that systems optimised at about 80V would achieve the best cost-effectivenesswithout introducing any serious hazard. Generally, it seemsthat there is a need for manufacturers to pay more attention to detail and to increase the level of quality control, inspection and pre-despatch testing that seemsto have been applied to most of the systems purchased during the course of the Project. 3.4 llre Importance of Local Manufacture An important consideration, if small-scale solar pumps are to be used in large numbers, is their suitability for local manufacture or at least part-manufacture and local assembly. This is because one of the primary motivating forces in seeking 56 alternatives to sngines for water pumping stems from the high cost of importing oil and the difficulty many developing countries are having with their balance of payments due to the oil price increasesof recent years. Clearly if solar pumps are imported, then similar problems arise in finding the foreign currency to pay for them as for importing oil. However, if a significant proportion of the capital cost of solar pumps originates within the local economy, then they would be attractive in terms of import substitution and reduced foreign currency requirements. Additional bene!i;s trotn local manufacture would be: 0 shorter supply lines from manufacturer to user, resulting in: *) b) C) d) more accurate system specification to suit local conditions reduced shipping costs minimisation of import formalities, and duties and the costs of intermediaries more rapid and responsive spare-parts availability and aftersalesservice creation within the local economy 0 job 0 enhancement and upgrading of technical skills and capabilities within the country 0 reduced dependence on imported fuels (and a greater level of national self-reliance) 0 if lower pumping costs are achieved, then impoved agricultural productivity and cheaper food may result with allied economic benefit to the rural population Therefore, the question of local manufacture or part-manufacture is important when considering the merits of specific small-scalesolar pumping systems for use in developing countries. It can be,expected that there will be a number of barriers to the development of the technology relating to the local manufacture requirement. A key one is that the large international manufacturers who might be expected to take an initiative in commercialising the technology are inhibited by a number of factors, such as:0 the potential market, although extremely large, consists mainly of poor farmers lacking capital resources.Therefore even when the technology is shown to be economic, the eventual market size will be dependant on intervention by governments and/or agricultural credit banks to set up suitable financing facilities. The consequent need for independent assurance is one of the main reasons why the UNDP/World Bank programme is of such importance. 57 0 unit costs, which are a key factor in achieving economic viability, are closely related to the scale of manufacture (particularly for photovoltaic c&s). A commitment to a large scale of manufacture (a large investment) or heavy subsidies appear to be essential at an early stage to bring system costs down. 0 decentralised manufacture within individual developing countries, with minimised external costs to the local economy, is against the shorter term commercial interests of many large transnational commerci*l interests, since such companies would seek a good return in hard currency. Thus some companies capable of promoting the technology will have short term constraints which effectively militate against such development. 0 the engineering industries in developing countries, on the other hand, generally lack the research and development capability to develop new products. Sometimes they actually have the capability but simply lack experience and hence the confidence necessary to embark on technical innovation. Researchinstitutions, many of which are associated with universities or government, often have the capabilities but lack adequate contact with the commercial secmr to convert experimental prototypes effectively into commercialiy viab!e products. There is a school of thought that technology that is both technicaUy and economically feasible will develop in responseto market pressures.However, there is no real evidence that this assumption always holds true -many potentially valuable inventions have not been exploited and developed. Also technical solutions are urgently required to the anticipated demand for a future expansion of food production which demands a strategic rather than an ad hoc approach from international agenciesand governments. If solar pumps appear to be essential to the improvement of food production, then resources for their future development may have to be provided from develop ment funds. There are many pressing demands for such Lids, so any large investment should not be made until it is clear that solar pumps are more worthwhile than other types of renewable energy pumping system. It was beyond the scope of this phase of the Project to make any indepth comparison between solar and other renewable energy resources for pumping, but it is hoped to look at this question in future phases. 58 4. ‘.r ..w~..“..T,. SGLAE PLmF~vu ~~&If+-GLGGY - PXOTOVOLTAIC SYSTEMS 4.1 Photovolt& Cells 4.1.1 Basic Technology Converting solar radiation directly to electricity is a comparatively new technology which grew from the semi-conductor revolution in electronics of the 1950’s. Photovoltaics were initially extremely expensive for power applications and their use was restricted to powering space satellites. However during the late 1960’s, improvements in manufacturing techniques combined with increased manufacturing volumes led to a steady decline in their unit cost and they began to be used for premium terrestrial applications such as powering remote navigational equipment, telecommunication repeater stations and, more recently, for isolated equipment such as railway signals and country telephone boxes. There are now at least 15 manufacturers of silicon solar cells, working in more than 9 countries. The only type of photovoltaic power cell commercially available so far is the sillcon cell, usually in the mono-crystalline form but recently also in polycrystalline form. Before discussing these and their costs in more detail, it is worth outlining the terminology used. The mono-crystalline silicon solar ceU itself consists of a thin slice of highly purified and then chemically doped crystalline silicon (i.e. cut from a single crystal grown from molten pure silico,r). Modem commercial cells are generally 1OOmm(4 in) in diameter (some are cut square or hexagonal to allow closer packing). Metal contacts are applied to the front and rear surfaces. When exposed to sunlight silicon cells develop a potential of 0.6V (at 250) on open circuit and will deliver around 25mA of current for each square cm of ceUsurface under an irradiance of I kW/ms ., i.e. giving about 2A from a single IOOmm diameter cell at OSV, which is 1W power rating. Cell efticiencies are typically 10 to 12%at present; i.e. a yield of 100 to 12OWcan be expected from every square metre of cell area in an irradian;:ce2.c :*‘-z”,-3. n .- : ... Poly-crystalline solar cells are similar except the wafers are cut from cast ignots of pure molten silicon, usually 1OOmm square in section. Their efficiency is typically about 9 to 10%. Cells are usually connected in a series“string” (making the cell voltages additive), the string of cells being built into a “module”. The module is usually a rectangular panel consisting of a glass (or sometimes plastic) window and a rigid back with the cells laminated between them. The cells are generally encapsulated in a suitable polymerised pottant to protect them from damage.Terminals for the cell string are provided externally on each module. A set of modules is then built into an “array”, which generally consists of a frame to carry the modules securely, plus a wiring harness and junction box to allow the modules to be connected electrically in a suitable series/parallel combination to suit the load, and to provide an off-take for their power. 30.0( Current Silicon Cell Technology -G $ IO.OC : 2 z Jet Proputston Laboratorie IUSAl Block Buys 5.oc 5 z 3 . 2.50 a. !! ;r . 1.00 $ -. Development-Automation-Scale - up 0~50 E” 0.25 Production Experience Thin Films and Advance 0. IO 76 77 78 79 FIGURE 12 - PHOTOVOLTAIC 60 61 82 MODULE/ARRAY 83 64 65 66 87 PRICE GOALS AND HISTORY (IN 1980 S) 66 69 so Calend.ar Year 60 4.1.2 C&S The cost tigures which follow relate to the large quantity purchase of modules (i.e. packaged cells) in 1980 US dollars. Power outputs are germrally given for cells exposed to lkW/ma of sunlight when at a temperature of 25°C. This is not a realistic operating condition, as they norm&y run sttbsiantirdly hotte.I than this which results in reduced performance, but it is an industry standard that coincides with solar cell testing techniques. The selling price of silicon PV modules has dropped quite dramatically in recent years; from S 2OOOjWpkin 19% (when their use for powering space satellites started), to S 30/Wpk in 1976 (for terrestrial modules), falling to S 1 l/Wpk in 1979 and to about S 9/Wpk in 1980 (from the manufacturer with the largest share of the market). Figure 12 illustrates PV prices since 1976 and some predictions for the future. This diagram is based on the former US Department of Energy published price goais (e.g. Ref 25) which are now almost certainly, over optimistic. The Consultants believe there is some possibility of the price dropping to S7/W pk by 19834 and perhaps 1-2 S/W pk by 1990. The selling price of solar cells does not necessarily have any close connection with either production costs or their current economic competitiveness with other energy resources,since some manufacturers, particularly in the USA, have been benefiting from the policy of the American government prior to 1981 of encouraging large scale production of PV cells through bulk purchases and various financial incentives for potential end-users.It is widely assumedthat some of the lowest selling prices reflect manufacturers’ strategies of improving their share of the market and volume of sales by selling “loss leaders” now, with the expectation of reducing their production costs and seeking profitability later on. Hence the selling price of PV cells is significantly affected by the marketing strategies of those manufacturers holding the largest current share of the market, and the lossessome are prepared to accept. The trend over the last year has been for the price of alar cells to level off: the lowest prices in 1981, at around S9/Wpk were snnilar to those in 1980, which, allowing for inflation, implies a small decreasein real costs. The range of prices from different manufacturers has also narrowed. Changes of policy are being introduced by the new US Administration which may be causing manufacturers to rethink their pricing policies. There was a shortfall in the production of so!ar cells in 1980 when the price reached a low level, possibly due to higher than anticipated demand, which may have also influenced pricing policy. Current mono-crystalline silicon cell technology still relies on a considerable element of hand assembly. It is expected that if there is an increase in the present market size, automated production will result in a further growth coupled with, and dependent on, further price drops. The most expensive item in the production of mono-crystalline silicon solar cells at present is the pure silicon raw material, currently costing well over $2/W pk (according to Ref. 27); improved processes,particularly the development of cheaper solar grade silicon and cast ignot techniques coupled with increased production should result in continued price reductions. 250 TABLE 7 - EXISTING SMALL-SCALE (Sheet 1 of 3) PUMPING INSTALLATIONS *omvoLTNc POWER w 120 250 IS TABLE 7 - EXISTlNG SMALL-SCALE (Sheet 3 of 3) 1: 4 PUMPING INSTALLATIONS 64 It is expected that other PV cell technologies will be commercially used within the next two years; these include thin film cadmium/copper sulphide cells and amorphous silicon cells. Both of these are “behind schedule” if the claims of their adherents made only two or three years ago were believed; but there is good reason to believe that some of the technical problems facing these technologies are being overcome and that they may well be competitive with crystalline silicon, but probably not until the end of the decade. In theory, thin film cells of these kinds could become significantly cheaper than crystalline silicon, once a high volume of production is achieved, so the price levels predicted in Figure 1’2for the cad of the decadr nii ,,;L:-. Although it appears that PV cells are still too expensive to allow solar pumps to be viable for irrigation, a further price reduction of 50% would make them potentially useful under carefully chosen conditions, while a 60-70% reduction would probably make them economically viable for very low head pumping (around 3m). Also, they probably are already viable for certain water supply duties in remote areasat moderate or low heads,but it was beyond the scope of the first phase of this Project to investigate this application. 4.2 Existing Photovoltaic Pumping Installations Unlike solar-thermal power systems, photovoltaics are now being widely installed in practical and economically viable applications, notably for remote telecommunications systems. There are currently at least 60 small photovoltaic water pumps in operation around the World. Most of these are being used for demonstration purposes and probably only those systems used for water supplies are economically viable as yet. The majority of these have been installed by the French company Pompes Guinard and the US company Solar Electric International (SEI). They all use flat plate silicon solar cell arrays but the type of motor/pump set and control system varies between the manufacturers and particular applications. Table 7 presents details of some existing installations. 4,3 Photovoltaic Pumping Systems The main components of a photovoltaic pumping system are illustrated schematically in Figure 13. Components indicated in solid outlined boxes are essential to any photovoltaic system, whereas the components indicated within broken line boxes are options that have been offered by some but not all system supplien. The energy flows (and losses) through a basic PV solar pumping system are shown in Figure 5. r- ----yrocking --w--B 1 Mechanism)(;- - 1 t- -----I Concentrator L -y--’ - d\ t- ----+ - - -I L Heot ----- 1 Sink \ \ Photovwltaic ---- --L \ 0 \ I \ / Motor I I L ----- Well L F::iot e-B-- Casing Valve Mechanical Transmlrrion 1 J 1 ---Jk --- -j r ----Reservoir L Plumbing ----- FIGURE13 -SCMEMATICARRANGEMENTOFAPROTOVOLTAlCSOLARPUMPINGSYSTEM 1 I J 66 Contact Grid ,n-Type A Negative Contact ‘Positive Silicon Contoci FIGURE 14 - SILICON SOLAR CELL Maximum Conversion lrradiance =I000 Cell Temperature 0.1 Note with 0.2 that the efficiency the Voltagepower Power Density Efficiency : 12.7 mW/cm? 01 0.46 V = 12.7% Wr?(lOOmWcm~) = 25O C 0.3 curve in this curve, since FIGURE 15 - VOLTAGE-CURRENT AND VOLTAGE-PObER SILICON SOLAR CELL 0.4 0.5 0.6 Voltage (VI case exactly coincides the input power is 100mW/cm2 CHARACTERISTICS OF A 61 4.4 Photovoltaic Amy 4.4.1 Monwrystalline Silicon Arrays The key component of any photovoltaic system is the array, which converts sunlight directly to direct-current electricity. The basic element is the solar cell, of which there are several types. However, the only type commercially available at present is the monocrystalline silicon cell, which has the considerable advantage of being basedon well-known semi-conductor technology. Its mode of operation is well understood and it has been manufactured in large quantities. One version is shown in Figure 14, consisting of a thin slice of specially-doped single-crystal silicon, commonly 50mm to !3Pmm in diameter or, more recently, square, with metal contacts on the front and back surfaces.The surface is often treated so as to minimise the amount of light reflected and thereby improve the efficiency. Figure 15 shows the voltage-current and voltage-power characteristics of a modem silicon ceUat 2SoC in sunlight at an irradiance of 1000W/m2 the so-called “peak” conditions. The short-circuit current (at zero voltage) is directly proportional to the cell area and the irradiance, while the opencircuit voltage (at zero current) is independent of cell area and is logarithmically related to the irradiance. The maximum power delivered is rep resented by the area of the Largestrectangle than can be fitted under the curve (being the product of voltage and current). In this case,maximum power of 12.7mW/cm2 is produced at a voltage of just under OSV. The maximum conversion efficiency is the maximum output power expressedas a percentage of the input power @radiance x cell area), in this case 12.7%. However, if the cell is operated at voltages away from about 0.45V. the power and the efficiency decline to zero at 0 or 0.6 V. The maximum operating voltage is normally fixed so as to be slightly less than the voltage for maximum power; i.e. just to the left of the “knee” of the curve. It should be noted that the current from a good cell is almost constant over the voltage range up to the operating point. Ar. increase in cell temperature causesa slight rise in short-circuit current Is, but a sharp fall in open-circuit voltage V,,, asshown on Figure 16. As a result the maximum power and efficiency fall by abwt 0.5% per degree c rise. Becauseof the loss of power at high temperatures, it is important in array design and installation to ensure that the cells run at as low a temperature as possible. Figure 17 shows the effect of a temperature rise from 25OC to 60°C on the voltage-current characteristic. The voltage for maximum power also falls with increasing temperature. However, providing the cell operating temoerature does not exceed that corresponding to the voltage for ma&mum power at the highest operational temperature, the output current is only slightly affected by changesof temperzture. 68 30 -600 _ 12 _ 20.400. 8 _ 01 - 40 I -20 0 lrradiance 20 =I000 40 Wni’ 60 Temperoture(‘C) 80 FIGURE 16 - DEPENDENCE OF EFFICIENCY, I,, AND V,, ON CELL TEMPERATURE Maximum Power Density = i2.7mW/cm2 lnadiance = 0 0 0.2 0.4 0.6 Voltage 0.8 (V.) FIGURE 17 - EFFECT OF CELL TEMPERATXJRE ON V-l CHARACTERISTIC 0 0.2 0.4 0.6 0.8 VOltoge ( v. 1 .FIGURE 18 - EFFECT OF CHANGE IN IRRADLUJCE ON Y-I CHARACTERISTIC 69 Figure 18 shows how the characteristic changesin cloudy conditions, when the iradiance might fail to 200W/mr. Power at the operating voltage faiis roughly in proportion to the irradiance but is still available at this voltage down to a very low light threshold. This is an important feature of photovoltaics. in concentrated sunlight, the short-circuit current increases proportionally to the itradiance up to extremely high levels (providing the cell temperature is kept constant) and the open-circuit voltage increases logarithmically, so one might expect the conversion efficiency to in-.,.ove. in practice, however, the series resistance of the ceU has a progressively flattening effect on the voltage-current characteristic, so higher efficiencies are not achieved unless steps are taken in the ceUdesign to reduce series resistance. This feature and the provision of adequate cooling are the main technological problems in developing ceiis for high concentration. Silicon solar ce!is, in contrast to thermal collectors, respond very rapidly to changes of h-radiance, their time constant being about 20 microsecond. The photovoltaic module, the basic building block of a Uat plate array, consists of a number of interconnected ceils, usually in a single series string, encapsulated behind a transparent window to protect the fragile cells from mechanical damage and the weather. The number of cells in series is chosen to ensure that the module will produce power at the desired working voltage at the highest expected operational temperature. in most commercial designs, an element of redundancy is introduced in the inter-connections between cells so that a single broken or bad connection does not causea complete failure of the module. To achieve the long working life that is required, the module must be rugged, capable of resisting ultra-violet radiation, thermal cycling, thermal shock, moisture ingress, fungi growth, sand, dust, hail, salt spray, wind loading, wind-induced vibrations and rough handling. A number of special hazards are also likely in the environment expected for irrigation systems, for example obscuration and damage by bird guano and damage by aninials (goats have even been reported trying to eat the encapsulant), plus the possibility of theft or vandalism. Glass is usually used as a window material in current designsand synthetic resins such as silicon rubber or poiyvhryibutyrai (PVB) as encapsuiants. However, the Consultants’ experience in the field trials has shown glassto be vuhrerabie to breakage,both in transit and on site. Polymer&d resins and polycarbonate plastics have been used in the past, but proved to be excessively pervious to moisture and prone to deterioration from dust. Possibly a laminated glass and plastic window cover would offer the virtues of both materials without their disadvantages. As a certain amount of spacehas to be allowed between the ceils, the conversion efficiency of a module, basedon the grossirea it occupies in an array, is less than that of individual ceils. Figures for commercialiyavailable modules range at present from 5 to 8% for those embodying circular ceiis and 8 to 11% for those with square ceils. The trend is to 70 increase the packing density of modties by using squared off cells in order to reduce the size and the cost of modules of a given power rating. In general, PV array efficiencies are quoted as a function of complete array area rather than ceUarea, and supplier and designersshould always be careful to defme efficiencies in quoted specifications. Modules are rated in terms of “peak power”, that is to say, the power they should pmduce at a specified working voltage and temperature (usually 25oC) in sunlight at an it-radiance of lOOOW/ms Becauseof the diurnal cycle and variations in atmospheric attenuation due to weather conditions, the averageoutput is considerably less than the rated peak value. For instance in some parts of India, Australia, the Middle East and Africa, where the mean global hradiance averagedover 24 hours is 250W/ma, a module rated at 10 peak watts could in theory produce 2.5W on average,or 60Wh/day. However in practice the operating temperature will be higher than 25OC and wili result in a reduction of output of typically 7’S% at 40°C. in addition, attenuating effects due to dirt accumulation and ageing of the encapsulation and window surface may further reduce output by 5 to 10%. Hence a module rated at 1OWpeak may only yield say 50Wh/day rather than the 60W:l/day predicted on the basis of ideal power ratings. The above analysis assumesthat the module is always loaded electrically so as to make it operate at, (or near to) its maximum power and maximum efficiency point. Failure to do this would reduce the daily output further. A further problem with PV atrays is bad matching of either modules or cells. inconsistent manufacturing can cause the characteristics of individual cells to vary. A sub-standard ceil or module in a series string may, under certain load conditions, be driven into reversebias, causing power to be dissipated in the affected element. in extreme casesthis can causeover heating and eventual failure. A similar effect can be caused by a fault such as a cracked ceil or by the shadowing of certain ceils or modules. With badly-matched modules in parallel, certain load conditions can cause a reversecurrent to flow through modules with low open circuit voltage, Ai<&,.dd *gdrii i*usL1; >Z”=;-- 1<.-.LL-7.>l;y...ec_. Therefore manufacturers are trying to improve the consistency of their products and to identify modules with slight variations so that they can be matched with similar ones when assembling an array. it is possible that, in future, massproduction wiU result in more consistent products from the industry than at present. Field experience with photovoltaic arrays in a variety of applications extends over a number of years. One French installation powering a copper electrolysis plant in Chile has been working satisfactorily for 20 years and photovoitaics have been used for educational TV in Niger over the past 10 years. in general, the arrays have functioned with a high degree of reliability and a life of 20 years seemsa reasonable goal. CIaims to have achieved this goal already with currently manufactured modules are basedon projections from accelerated ageing tests rather than on real experience. During the laboratory tests conducted in PhaseI of the Project, it was found that although the modules tested were adequately reliable in that they almost all sunrived the testing programme, they ah had a number of manufacturing defects. Most of these were minor, but a few were potentially serious. There was some variability in the performance of the modules tested and most of the outputs and efficiencies were below their manufacturers’ specification by from 2.5% to 16.5%. Clearly, the great virtue of flat plate photovoltaic arrays is their ability to generate electricity from the sunlight without moving mechanical parts. The resulting reliability and easeof maintenance will be important considerations in the preserrt application. The present high cost of photovoltaic arrays is mainly due to the expensive single crystal silicon and the use of labour-intensive batch production methods in the fabrication of cells and modules. However, extensive efforts are being made to reduce the cost of the starting material and to develop automated production processesas the market expands. The US Department of Energy had agreed to belo fund a $6M plant which will use Union Carbide’s silane process to produce about 10: tooF.?sof pure silicon per year and which, it is claimed, should bring the cost of silicon suitable for photovoltaic cells down from well c~;si $2.00 to around SO.50 per peak:watt. Other approachesto cheaper silicon are the heat exchanger method (HEM) being developed by Crystal Systems,USA, cast polycrystalline ingots (Solarex, USA, and WackerChemitronic, West Germany), continuous ribbon growth (Mobil-Tyco, USA, and Japan Solar Energy Co., Toshiba 2nd Toyo Silicon Co., Japan), dendritic web silicon (Westinghouse, USA), zone growth by laser beam melting (Motorola, USA), and sheet growth on low-cost substrates (LEP, France and others). The Stanford Research Institute is also understood to have achieved a breakthrough by purifying silicon in one step instead of two, thereby reducing the cost of pure silicon from $60/kg to under S lo/kg. 4.4.2 Other Types of Photovoltaic Cells under Development Another approach to cost reduction is to replace the silicon ceU by a potentially cheaper type. The main contenders are as follows: (a) Cuprous Sulphide/Cadmium Sulphide This solar cell consists of a p-n junction between two different semiconductors: n-type cadmium sulphide and p-type cuprous sulphide, most of the light being absorbed by the latter. In the “backwall” type, the tight enters through the cadmium sulphide, which is deposited on a transparent substrate. In the more common “frontwall” type, the cuprous sulphide layer is illumin- 72 ated. Figure 19 shows the cross-section of a typical frontwall thin-film cell, which was originaily developed for space applications and is now being adapted for terrestrial use. This cell is potentially cheaper to make than mono-crystalline silicon types and its manufacturing processesare adaptable to mass production. tiui ii is not so well ur,dsrstood and its technology not so well advanced, despite the large sum of money spent on its development in USA, France and other countries. Efficiencies have not advanced beyond the 9% mark and are commonly below 6%. The theo:e:ica! maximum has been estimated to lie between 11 and 14%. Cu2S/CdS cells are not yet as stable or as reproducible assilicon cells. The main cause of degradation is the oxidation of the cuprous sulphide to the less efficient cupric sulphide. To prevent this, hermetic encapsulation must be employed. Pilot production lines have been set up in USA by SES Inc. (frontwall) and Photon Power (backwall on glass) but neither company has yet been able to offer modules on the market, although SES Inc. claim that they are close to commercialisation. Despite the difficulties, there is still a considerable body of opinion which regards this type as the main contender in the search for a really cheap solar cell. (b) Schottky Barrier A metal/semiconductor junction (Schottky barrier) has photovoltaic properties similar to those of a p-n junction and solar cells embodying such junctions can theoretically be made as efficient. This type is suitable for low-cost massproduction. The metal can be either in the form of a substrate supporting a thin film of polycrystalline or amorphous semiconductor material or a thin transparent layer over the semiconductor. The cross-s&ion in Figure 20 shows the essential features of the latter version. Research on Schottky barrier cells is in the early stagesat present, although an efficiency of 11.7% has been reported with single crystal silicon as the semiconductor and a possible 15% with gallium arsenide. A key factor, and one that it is proving difficult to control, is the thin oxide layer which it has been found necessary to interpose between the metal and semiconductor to raise the open-circuit voltage to an acceptable level. Because of this insulating layer, some investigators are applying the term “MIS” (metal-insulator-semiconductor) or “AMOS” (anti-reflectivemetal-oxide-semiconductor) to this type of cell. The operation of Schottky barrier cells is not yet well understood. Cc) Amorphous Silicon The optical properties of thin films of hydrogenated amorphous silicon are such that it should be possible to make efficient solar cell from a thickness of only 1 or 2 ,um. Since crystalline si!icon Cek require 300 to 400 @m,this represents a dramatic saving in Encapsulant K6pton Epoxy Cement Substrate Metallised FIGURE /Front Grid Contact 19 -CADMIUM Layer SULPHlDE Epoxy Cooled Myla SOLAR CELL Grid ,Front Contact Epiloxial I I Grid p-Type Epitaaiol p-Type Eoitaxiol Go,-~AI~As Window Ga As Doped) n-Tvoe (Zinc Go As 1//;//////////////////~ Go AL ’ Back FRXJRE 20 - SHOlTIKY ‘Ehck Contact BARRIER (MIS) SOLAR CELL FIGURE Substrate Contact 21 - GALLIJM ARSENIDE SOLAR CELL I 1 I 74 material. Both homojunction and Schottky barrier cells are being developed with amorphous silicon but the work is in its early stages and many problems remain to be overcome. Best performance to date was achieved by RCA, who made a Schottky barrier cell with an efficiency of 5.5%. Cd) Gallium Arsenide The modem gallium arsenide ceU consists of a thin epitaxial layer of gallium auuminium arsenide super-imposed on thin epitaxial layers of p and n-type galllum arsenide, all on a substrate of single crystal gallium arsenide. Figure 2! shows the construction in cross-section. Its main attraction lies in its ability to operate efficiently in highly-concentrated sunlight, which opens up possibilities for off-setting its cost, which is much higher than that of silicon and likely to remain so. Cells have been operated at concentrations of up to 5000 suns although a more practical limit is 2000. Efficiencies of over 20% have been claimed. However, production has been only on a laboratory scale so far. 4.4.3 Concentrating photovoltaic arrays Photocells arr perhaps the most expr.*,ne >Angleelement in a PV pumping system. Sir-e their output can be b ,osted simply by increasing the intensity of illumination, it is possible to reduce the area of cells necessary for a given power output (and their cost) by concentrating sunlight onto them. Reduced overall costs have been claimed for this becausethe extra sunlight concentrating equipment can cost less than the extra number of cells that would be needed to provide the same output without concentration. Methods of concentrating the sunlight that are commonly used include focussing mirrors or using Fresnel lenses(usually moulded in plastic). Figure 22 shows the principal methods schematically; the cells can be between or surrounded by reflecting surfaces (top diagram), at the focus of a parabolic reflector, facing away from the sun towards the reflector (centre diagram), or behind a lens, at its focus, (lower diagram). In the latter casea Fresnel lens is usuaUy used as this can be kept quite thin and light in weight and can be massproduced cheaply from clear plastic. By contrast, a normal convex lens would have to be very large and expensive for the apertures involved. Figure 22 is only two dimensional. In practice any of the concentrating systems shown can be either point focussing or line focussing. In the former case the reflector or Fresnel lens is normally circular (or polygonal) in aperture with the PV cells on its central axis. In the latter case the cross sections illustrated would extend linearly with the cells mounted as indicated as a long narrow strip. The degree of concentration is related to the ratio of the aperture and the area of cells at its focus; so-called “point” focussing results geaerally in a higher concentration factor, the image being small in area rather than a point. \ ‘\\ ,- Fresnsl lens \ FIGURE 22 - METIIODS OF CONCENTRATING SUNLIGHT ON PHOTOVOLTAIC CELLS 76 Although concentrating systems might save on cell costs, they have a number of disadvantages,which are as follows:(a) Concentrating the sunlight causesthe cells to heat up (to potentiaiiy very high temperatures at high concentration ratios) and hence cooling of the celis becomesnecessary.This can be done passively using air cooled fms as a heat sink (as illustrated) for low concentration ratios, but water cooling is usually necessaryfor high ratios. Any faihue of the cooling could result in destruction of ihe expensive cells. (b) Most types of concentrators need to be tracked in order to keep them correctly aligned with the sun. The higher the concentration ratio, the more accurately must they be tracked; (some very low concentration ratio systems need only be tracked very occasionally suet as every few weeks to cater for seasonal changesin the sun’s course acrossthe sky). Tracking introduces complication in the form of a mechanism (clockwork, electrical, or powered by solar-heated vapour) and it not only requires the array to move readily in responseto the tracker, but also to be rigidly located so as to resist any movement due to wind gusts; hence an expensive support structure may be needed. (cl Concentrators can only focus direct beam radiation from the sun, so the diffuse component cannot be used. Therefore concentrators are only likely to be effective in areaswith arid climates having a large proportion of direct sunshine with limited haze and cloud. In comparison with fixed flat-plate colhsctors, tbis shortcoming is partially compensated for by the fact that in tracking the sun more energy can be collected, as described in section 2.7 and illustrated in Figures 4A and 48. (d) Uneven iRumination of the celis is a problem, particciark~ with low concentration ratio systemswhen celi strings s;c >I parallel aswell as in series, as this can causeinternally ~ircuiating currents and “hot spots”. Uneven illumination can be causedby blemishes or discontinuities in the reflector(s), or by a non-continuously tracked system being misaligned with the incoming beam radiation. (e) Reflectors (or lenses) involve a loss due to imperfect reflectivity (or transmissivity); this is at least 10% of the energy flux per reflection (or transmission) and possibly substantially more. A major problem with reflectors, panictiilariy in arid areas where the high proportion of direct to diffuse radiation is to be found, is degradation, either temporarily or permanently, due to dust collecting on the reflecting surfaces.With flat plate arrays this is less of a prob!em as although the dust scatters the light, much of it still is received by the cehs. When being compared to a Rat photovoltaic array, a concentrating wray with its associatedtracking system can only be justified if the cost sa+ng from using fewer solar cells ourweighs the extra cost and complication of 77 tracking, account being taken of the extra energy gained by tracking and the energy lost through inability to utilise the diffuse component. It should be t.orne in mind that flat-plate arrays can benefit considerably from occasonal tracking, but do not need the expense and complication of continuous tracking. The benefits expected from this and other options are illustrated by Figures 5A and ZB and are quantified later in this chapter. As the cost of PV cells falls, the balance should move in favour of nonconcentrating arrays, especially for small systems.Present lowest costs per peak watt .:laimed by manufacturers are similar for concentrating and non-concentrating arrays This suggeststhat if the price of PV cells falls substantially (which it must do if solar irrigation pumping is to become viable) then, since the price of mirror reflectors and tracking ;. unlikely to fall as math az the price of cells, it seemsunlikely that concentrating systemswill prove competitive ln the long run. There are two configurations which might prove to be cost effective. 0 very low concentration by adding small reflectors to the periphery of a llat plate array might well be cost-effective due to the simplicity of such an arrangement and rhe fact that tracking would not be necessary and much of the diffuse component would still be useful. This implies a reflector area probably less than that of the cells, simply used as a booster. 0 it is possible that systems with very high concentration ratios might be cost-effect&e through being able to use rather small areasof high cost but high efficiency photovoltaics (such as, for example, gallium arsenide) which would in turn possibly allow a significant reduction in the array aperture required for a given electrical power output. This would be likely to be cost-effective only if substantial gains in celt efficiency prove possible due, to the various concentrator lossesthey must outweigh. However, if gross array efficiencies o: over 20% or so prove possible through this approach, it would probably be worthy of serious consideration. As discussedin Chapter 2, the sizing of a small-scaleirrigation pump is critically dependent on the peak irrigation water demand and the solar energy availability in that month -for most months the system will be oversized for the actual water needs.Therefore, manual tracking of the array may be important to minimise this mismatch, since careful tracking during times of peak water demand will go some way towards reducing the required system size and cost while at other times of the year less careful or no tracking may be adequate. It is open to question whether the extra costs and continuous complication of mechanical automatic tracking systems will be justified, when the benefits will probably only be necessaryin the irrigation application during the one month of maximum irrigation water demand. Further investigation of automatic tracking system costs nnd the actual capabilities of farmers at manual tracking will be essenr~l to clarify this iSSW. 78 4.5 ElectricMotors Most system suppliers at present favour dc permanent magnet motors. This type of motor, although more expensive than field wound dc motors, offers such a good efficiency, particularly under part-load conditions, that it is virtually the only sensible option. As described in the Pro+ct Report, most of the motors tested were conventional with a segmented commutator and brushes. Such machines generally require new brushes at intervals of the order of 2000 - 4000 hours, and if this is not done, certain models can suffer irreparable damage. Some dc motors are being offered with claimed brush lies of about 10000 hours, and these would be better for this kind of application if these lives were acbievTedunder Reid conditions. With any brushed machine, fail-safe brush design is essential, so that the machine will stop once the brushes are too worn. rather than damageitself. No investigation of actual brush life was possible under PhaseI of the Project. Another problem with brushed motors is the build up of carbon dust that arises from wear of the brushes, which can cause arcing or overheating of the armature or premature wear of the bearings. An alternative type of permanent magnet motor (one of which was tested during the fmt phase of this Project) is brushless with the magnets in the rotor and an electronically commutated stator. In principle, such machines seemmore attractive than brushed machines for small-scale solar pumping, since their only wearing parts are the bearings and any seals. However, although early problems with the electronics have been corrected, the electronic circuiting is still vulnerable to overheating due to the high temperatures that can be reached when the motor runs for long periods, especially in the tropics. Thus they need effective protection from damage by high current or voltage transients, from external damage (eg by humidity), and from overheating through good heat sink design. The electronically commutated motor which was tested was slightly less efficient than the best of the brushed dc machines tested. It is likely that the higher power drain from the electronic commutator is inherently the reason for lower efficiency although electronically commutated motors could probably be built which show a less marked drop in efficiency compared with the best brushed machines. Brushes are a source of energy loss due to the resistive lossesthey imply and due to friction with the commutator surface, all of which can be set against the electronic commutation circuit’s parasitic power dram. However, a higher efficiency electronically-commutated motor is likely to cost more to manufacture. The laboratory testing programme completed during the Phase I of the Project indicated that the better dc permanent magnet motors, in the 250 to SOOWrange are capable of optimum efficiencies of around 85% and even at half load their effciencies should exceed 75%. A more detailed description of the testing programme is included in the companion Project Report, but Figure 23 indicates a typical dc permanent magnet motor characteristic obtained during the testing programme. Some of the poorer electric motors were only 75 - 80% efficient. However it is worth noting that an 85% efficient motor requires 9% less array area than one of say 78% efficiency - with a typical 400W array this translates into a capital cost saving at the array of typically one 36W module, worth about 6360 at present day prices. 2+eM cmtwI h rpm -_---*Eftlclmy conlw6 in *A. -_-Taw conhwm in Nm. N= [email protected]&6 T = -8.45 &4V+0.1361-O~0674 .5- oL---. 0 5 IO 15 m #.. . I.. Il.. 40’.. ‘45’.‘&O’. .&---?. 25 30 3.5 FIGURE 23 - TYPICAL DC PERMANENT MAGNET MOTOR PERFORMANCE L Mllac4 w a0 Mass produced ac motors are about half as efficient as dc permanent magnet motors in the small power sizes of interest. In addition, they require the use of an mverter (to convert dc to ac) and, usually, a battery too, both of which imply further losses in effkiency. They are generally inefticient as they were not designed with efficiency as a high priority, (the costs of the electricity used in mains applications for fractional horspower motors are small) and low unit costs are the main preoccupation of manufacturers of such equipment. Therefore, although they might be half the cost of a dc permanent magnet motor of the same power rating, the lower efticiency will considerably outweigh this benefit I hrough the increased array size required. A linear actuator was also tested - this is a reciprocating solenoid device, with permanent magnets, but brushless. The assumption was that this might be both efficient, low in cost and simple if coupled directly to a reciprocating pump. !.ack of a suitable pump hampered this part of the programme as it became clear that any pump wouid have to be carefully matched to the actuator so as to achieve a good efficiency at the frequency of vibration used by the actuatcr. The actuator was designed to operate from 50Hz ac supplies; therefore an inverter or other switching devices would be required to generate a suitable ac current from the dc output of an array. Further investigation might be justified since devices of this kind should be inexpensive and long-lasting a!thougb it is not clear whether good efticiency can be achieved in practice from an actuator-pump combination, The main advantage Of using a battery storage in the system, regardless of whether it delivers current via an inverter to an ac motor or direct to a dc motor, is that it can provide a steady electrical output even if the photovoltaic array output tkctuates considerably due to the continuous variation of angle of received radiation and to passing clouds. It also offers a reserve of power to provide the surge of current needed to start most electric motors and pumps, a particularly useful feature if the pump has a high breakway torque, such as with some positive displacement pumps. In virtuakly all casesa control system is necessary to regulate the battery charging process and to switch on the motor/pump combination when the battery stateofeharge is adequate and to switch it off when the charge level falls too low. Batteries however involve a number of significant disadvantages, namely: 0 they are an expensive component; 0 they getter& need regular topping up with distilled water (sealed mamtenance-free batteries are availabie, but are considerabiy more expensive). they give rise to a significant energy loss as the charge-discharge cycle will generally be in the range 50 to 80% efficient, which necessitates a larger photovoltaic array; and 81 0 the life of conventional lead/acid batteries is limited to about 5 years under field conditions, considerably less than the desirable life of the system as a whole. It is worth noting that the principal photovoltaic system manufacturers, with experience of solar pumping in developing countries, offer systems without batteries, the dc motors being powered directly from the photovoltaic array. It is not expected that batteries are likely to be cost-effective for sma&scaIe solar pumping systems used for irrigation at low heads, where the pumps need not have a high starting torque. 4.7 Pomps 4.7.1 GeneraI In the Consultants’ view, an essential requirement for any solar pumping system is the use of a pump that will reliably self-prime, even if the foot-valve (where fitted) is imperfect. A large proportion of the low head systems currently commercially available do not meet this requirement. The self-priming requirement narrows the choice to: 0 submersible centrifugal combined motor/pump units. 0 a centrifugal sump-pump arrangement, with the pump below water, driven by a shaft from a motor above the water 0 self-priming surface-suction centrifugal pumps, with a priming reservoir to keep the impeller and casing flooded even in the event of foot-valve failure. 0 positive-displacement pumps which are inherently self-priming, (eg. piston, plunger, diaphragm or progressive cavity) However, for the low heads that are appropriate for small-scale irrigation positive displacement pumps seem inappropriate on grounds of efficiency, size and hence cost. 4.7.2 Centrifugal pumps The conclusions from Phase I of the Project were that a simple single-stage centrifugal pdmp appears to have the merit of adequate performance potential combmed with compactness and simplicity as well as having low starting torque requirements. Centrifugal pumps also offer the possibility of achieving a close natural match with a PV array (without recourse to power conditioning) over a broad selection of operating conditions. The importance of power conditioning is explained later under the section on PV System Design. .* -.----- . . I.5 ’ ?A. . 2~0n . . . 2.5’ P- Flow rate (I/s1 FIGURE 24 - TYPICAL CENTRIFUGAL PUMP PERFORMANCE Speed corhm in rpm Elficiency contwrslnO, hpul power cal,ouTr in Watts It-TabI [email protected] Ml) Imad 20 -.-- 15 I ( 0,...,,.... 0 05 . . . ..I.. IO 15 1 . . PUMP PERFORMANCE . . 2.5 O-Flow FIGURE 25 -CENTRIFUGAL . 20 rote I I/s) WITH FLAT SPEED CHARACTERISTICS - -- Speed contars in rpr Eifriency contours In’ hpul parer conlolls in Watts H-Total pwn~npadtmad (ml 20 -_-- 0 0 05 IO I5 2.5 20 0- Flow role (1:s) FIGURE 24 - REGENERATIVE COKTRIFUGAL PUMP PERFORMANCE --- Speed contours in r p m Efficiency cantours in 01 hpul power iwllLnln in Walls -_----- Spudcordarrinrpm Etlicincy cwdwrh9 kul p0w.r cultin Wottr I. Speeds shown are pump speeds. For motor sped, rmlliply by 86. 2. Etficiencbs itida belt transmission bo1wm-o motor and pnnp. o-l....,,...,... 0 ., OS I.0 . I.5 . . . . 20 "I ' I 2.5 O- Flow rate (I/B) FIGURE 27 - WFICAL POSITIVE DISPLACEMENT PUMP PERFORMANCE Ii-TohiI punuled WI) head 20 1 -.-_--_ Manufacbrsr’s 0 05 1~0 I.5 2,o FIGURE 28 - TYPICAL ROTARY POSITNE DISPLACEMENT data for 720rpm 2.5 0- Flow role PUMP PERFORMANCE Speed conth~~ in rpn Efiiciency contours In’ hlpll porrer cnntoln in Walls ( !A) H-Total punvad h) had 20 ---I I5 1 I IO- 5- 00 0 05 I.0 I.5 20 2.5 Q- FIOW rate (I/s) FIGURE 29 - FR1.F I) .Pli :~4GM PUMP PERFORMANCE -., --- Speed cordws in rps Etficimy conlwrs In hput pmer CQllrn in Walls “‘,,”:’ 88 An interesting example of rotodynamic pump is the regenerative [email protected], sometimes known as a side-chamberpump. In this pump, the water makes more than one transit through the impeller by being redirected back into the impeller by passagesin the casing. The entire inplIer does not therefore run full and lossesdue to water leaking back from the high pressure to the low pressure end are reduced through maintaining smaUcleuances between the impeller and its casing. It has the advantageof being iolerant of considerable speedreduction, and like most cent&[email protected] pomps, it ought to have low starting torque requirements. It can also tolerate a wide variation in head compared witb most centrifugal pumpr A performance characteristic for a pump of this type is shown in Fii 26. However, the example tested was of poor efficiency compared with the better centrifugaI pomps and was prone to suffer from binding between the impeller and its casingdue to the small clearances.It seems Iikely that these disadvantagesare inherent in this type of pump and may make it unsuitable for solar pumping systems where high efficiency is important. Also there is often a need to pump water containing suspended solids, which makes smaUclearancesa serious disadvantage. 4.7.3 Positive Displacement Pumps Positive displacement pumps can be more efficient than centrifugal pumps, but generally only at higher heads.The characteristic of a typical example is shown in Figure 27, which illustrates some of the differences. It has vertical, rather than horizontal constant speedlines, becauseflow with a positive displacement pump is essentially a function of speed (whereas head is a function of speed squared with a centrifugal pump). IncreasIng the head with the positive displacement piston pump increases the power demand and the efficiency, becausefrictional forces become smaller relative to the hydrostatic forces at higher heads. Since PhaseI of this Project was only concerned with low head pumping, positive dir placement pumps under optimum conditions of headsin excessof I Sm. were not tested. Therefore, positive displacement pumps of this kind are unlikely to be of much relevance for irrigation applications, but may have an important role for solar water supply pumps. One type of positive displacement pump is the surface suction pump. A problem associatedwith this pump is that it can only be used down a large diameter weU (on suitable supporting steelwork), or in situations where it can be located within about 5 or 6m of the water level. An example of a borehole positive displacement pump, which can operate submergedwith zero suction lift is the progressivecavity pump. This type of pump is one of the commercially-available rotary positive displacement pomps and has a good reputation for its tolerance of aggressivefluids (be they abrasive suspensionsor chemically corrosive) due to its rubber stator. Figore 28 illustrates the characteristic of an example of this type of pump, although the particular model tested was not adequately efficient for headsof less than 1511to compete with centrifugal pumps, but may well be suitable for use with solar water supply pumps sopplied from boreholes. 09 Pump efftciency is critical to the cost-effectiveness of the system and high efficiencies require good design with such small pumps. The indications from OUTtest pmgrammes are that an optimum pump efficiency for smaUpomps (200 to SOOWshaft power requirement) for use at low heads (in the 5 to IOm range), of around 55% is realisable and that consistent efficiency of over 45% at reduced speedsor non-optimum heads should also be achieved. Some systemstested had pumps only half as efficient as this, which effectively doubles the cost of water from them. What is not yet known with any certainty is the trade-off between high efficiency and the need for an impeller with a long life and a good tolerance of aggressiveimpurities in the water; good efficiency CM be obtained with narrow passagesand small clearanceswhich are of Course undesirable. Siarly, good efficiency can also be bought with a high speedimpeller, which is again counterproductive from the point of view of achieving the longest possible operating life and generouswater passage dimensions. Further longer term investigation of pump durability will be necessary to arrive at conlident conclusions on the likely performance Limits that might reasonably be expected. A typical pump performance characteristic produced during the Project test programme is given in Figure 24. The sensitivity of this kind of pump to variations in head is clearly shown by the efticiency contours in the diagram. This particular pump, which was one of the inure efficient ones tested, was unusual in having quite steeply sloping constant speedcontour curves, implying that it is tolerant of a significant percentage reduction in speedbefore it ceasesto deliver water at a given head. Most of the pumps tested had flatter speed contours more like the one whose characteristic is shown in Figure 25. Flat speedcurves, imply that a small percentage speed reduction will causedelivery of water to cease.However this was one of the most efficient small centrifugal pumps tested, achieving over 24% efficiency at a remarkably low rotational speed. Off-the-shelf industrial centrifugal pumps, as are commonly used for smail-scalesolar pumping systemsat present, are generally designed for single speed operation when driven by mains electric motors and hence many are intolerant of any significant speedreduction below the speed for OptimUm efficiency. Purpose-designed pumps for small-scale solar pumping will no doubt allow significant improvements to be obtained which will be measurable in terms of array size reduction through improved efficiency, as well as greater tolerance of speed and head variation. Becausepomps are less efficient than motors, a ten per cent absolute improvement in a pump from 40% to 50% efficiency results in a 25% reduction in array costs. At present array costs, a one per cent marginal change in pump efficiency, (with good pumps in the 40 - SO%efticiency level), would be worth $40 for a 200W (electrical) system and $80 for a 400W system in array savings. 90 One other generic type of pump tested was the free-diaphragm pump whose characteristics are illustrated in Figure 29. There was some hope at the outset of the testing programmes that diaphragm pumps might, through low internal friction, prove a superior type of positive displacement pump for low head applications. In the event, this particular type proved not to be competitive with the better centrifugals, although testing of one example of one type of pump genre cannot be conclusive. The reason for the lower efficiency than expected is not clear on the basis of the limited testing so far completed, and the original hypothesis has not been totally disaproved. The most weU known type of positive displacement pump was also tested -this was in the recipmcating piston pump or “bucket” pump of the kind commonly used as a hand pump. These are well understood and widely applied in agricultural applications, common examples being hand, diesel, electric or wind powered via suitable transmission systemsto reduce the input power source to a reciprocating drive in the range 50 to 100 strokes per minute, or sometimesmore. However they are not efficient at heads much below 10 or 2Om and therefore of little interest for irrigation duties. 4.7.4 Importance of Minimising Pipe Losses The system design study (see section 4.9) confvmed that lack of care in specifying the delivery pipework associatedwith the pump can seriously affect system efficiency, particularly with low static heads. This is particularly important in situations where a long pipe is involved. Relevant points are discussedin section 4.12. 4.8 Meehenical Tmnsmiasions Since PhaseI of the Project was concerned with low head pumping, transmission systemswere not considered in detail. A few general points are reviewed below. Most centrifugal pumps will be direct coupled to a motor, either by a long shaft, or in the caseof surface-suction pumps and integrated submersible motor-pump units, close coupled. This is because the speed requirements for economica! motors and for economical centrifugal pumps are similar, i.e. typically in the 2000 to 3000 rpm range. With direct-coupled pumps, it is advantageous(except with integrated motorpump units) if both motor and pump have completely independent shafts and bearings, (i.e. 2 bearings each) and if the coupling is flexib!e to allow for a certain degree of misalignment; otherwise premature bearing wear can be expected. Most positive displacement pumps run at much lower speedsthan motors (piston pumps typically at 50 to 150 strokes/minute) and a speed reducing transmission is necessary.But, as already explained, this type of pump is likely to be used for higher head pumping duties than would be appropriate for irrigation. 91 A crlticsl requirement from any transmission is that it should havehigh efficiency as weU as the necessaryrobustness and long life required from a solar pumping system. Possibilities for such duties inciude belts and pulleys (preferably poly-vee or toothed, although the latter can transmit damaging torsionai shocks from a reciprocating pump). Various types of gear box and low speed pitman drives would be appropriate for use with slow-speedreciprocating pumps located in deep boreholes. Photovohic Pumping System Optimisntion The characteristics of PV cells and arrays have been described earlier in section 4.4 To recapitulate; the characteristic of a PV array (such as illustrated in Figures 15, 17 and 18) is unique at a given b-radiance level and array temperature. This is further lllustoated by Figures 30a and b. As shown in Figure 15, the maximum efficiency of a ceU, or of an array of cells, occurs for the values of I and V on the characteristic such that the product of 1 and V is maxhnised. Since the product of I and V for any given point on the I-V curve is a rectangle of sides I by V, the maximum power point on the curve is that which produces a rectangle with the largest area capable of fitting within the curve and it will therefore be on the “knee” of the curve. Clearly it is important to operate any PV system so as to draw a current such that the array will operate at or close to its maximum power point. Figure 30a shows the maximum power point for three irradiance levels(500,750 and lOOOW/sq.m.)at a temperature of 40°C. A curve m-m can be drawn through these points which is the locus of the maximum power points (or maximum efficiency points) for that tempershire. It can be inferred from~Figure 30b that changesin temperature will tend to move the m-m locus slightly from side to side. The I and V values for the PV subsystem, or load, at any given instant must be the sameas those for the array, so the actual I-V valuesthat will occur for a given irrsdiance level depend on the impedance of the electrical load. (Impedance can be quantified as the ratio of V/I; with purely resistiveloads, the impedance equals the resistance,measured in ohms if V is in volts and I in amps). If the impedance is very high, such as on open circuit, I will be zero and V will be V,, (i.e. the point at which the I-V characteristic cuts the V axis) and if the impedance is very low, such as a short-circuit, V will be essentially zero and I will be Isc, (where the characteristic cuts the I axis). Figure 15 shows that the power output and hence the efticiency is effectively zero under the two extreme conditions mentioned. In order to maximise the output (and the efficiency) of the system under all operating levels of irradiance, the load should be such that its impedance should vary to coincide as closely as possible with the I-V ratios represented by the maximum power point locus m-m in Figure 30a. In practice, it is common for the load to be correctly matched for just one level of irradiance, such as say lOOOW/ m, but for the subsystem power demand at lower levels of b-radianceto cause the array to operate further and further from optimum such as illustrated by locii m, - m2 or m2 - m2 in Figure 30a. Figure 3Oc shows that, up to the maximum levels of voltage and current the motor can safebybear, there are unique values of torque and speedcorresponding to any unique value of voltage and current. If the array characteristic is super- 92 Total irradionct For 1000 W/m2 irradionce normal to orroy normal to array 1 i 750 W,m~ $, Array lem~erature 40-C V to) ARRAY CHARACTERISTIC WITH VARYING IRRADIANCE LEVELS (b) ARRAY CHARACTERISTIC WITH WITH VARYING TEMPERATURES Speed contours ‘N’ I, Torque Co”tours ‘T’ (c) 5 T4 MOTOR CHARACTERISTIC WITH MAXIMUM ARRAY POWER LOCUS SUPERIMPOSED Torque contours IT’ “I Unique operating pojtt 0 (d) ARRAY MAXIMUM POWER LOCUS TRANSPOSED ONTO PUMP CHARACTERISTIC (e) MAXIMUM POWER LOCUS SUPERIMPOSED ON PIPEWORK CHARACTERISTIC FIGURE 30 - PERFORMANCECHAXACTERISTICSOF PHOTOVOLTAICSYSTEMCOMPONENTS 93 imposedon the motor I-V plane,it generatesa unique set of speedand torque values,asdoesthe maximum power point locus m-m.i.e. for every point on the may maxbnmn power point line there will be a value of motor speed and toque +ich is required in order to draw the correct current and voltagefmm the array. If the motor is direct-coupledto the pump, then the motor and pump speedand torque at any givenmoment must be coincident: where the pump is driven via a transmission,a fixed ratio wiIl apply together with a smallpower low due to the transmission.Pump characteristicsare most conveniently representedon a head versustlow plane (HQ) asshown inFIgure 30d. Any valueof head and flow for the pomp will require a unique pump speedandtorque, and thesecan be superiztpused on the H-Q plane asa set of constant speedand constant torque curvesasshown in the Figure. The torque and speedvaluescoinciding with the array optimum power line m-m in Figure 30~ can then be transformedinto a coincident line on the pump HQ plane,asindicated in Figure 30d. In other words there is a unique set of head and tlow conditions which optimally loads the motor and in turn the array for all levelsof irradiance.Theseare indicated by the curvem-m in Figure 30d. The pomp is connectedto pipe-work which will havea characteristicin the H-Q plane of the kind shown io Figure 30e, consistingof the sum of the static headand pipe energylosses.If the curvem-m from the pump HQ plane is transposedonto the pipework H-Q plane,(since the head and flow experiencedby the pipework will coincide with those for the pump), then it cau be seenthat there is a unique singleIIQ point on the m-m systemmaximum power or efficiency curvewhere maximum efficiency is obtained from the system.This unique point can be transposedback through the variousplanesto the array and resultsin a unique iradiance level,(and array temperature),where maximum efficiency is obtained from the system.This showsthat for any PV pumping systemthere is a unique valueof pumping head(total headacrossthe pump) and flow which will optimally load the array at any irradiancelevelabovethe system’sthreshold. It follows from this that PV solar pumping systemsneedtheir componentsto be carefully matchedso that: 0 the motor-pump subsystemloads the array in such a way that as the irradiancevaries,the voltageand current requirementsremain at or close to the maximum power point locus. 0 the systemoptimum operating point, which will lie on this locus, should be correct for the headand flow required at the designirradiancelevel. Clearly the correct matching of subsystemsand arrays is a complex analytical processwhich is greatly facilitated through the useof a computer model which permits an iterative approachto be used. Figure 3 1 showsthe measuredcharacteristicsof a typical solar PV pumping system (PompesGuioard-boreholesystemat Yangassoin Mali), and it can be seenfrom the lowest curve in the Figure that this systemis optimised to achieveits maximum efficiency at between 700 and 800W/mz of irradiance.It can also be seenthat the systemrequiresabout 200W/m* to start and it achievesa maximum system efficiency of almost 3.5%. 94 95 Figure 32 shows the daily output of the samesystem recorded over a period of severalweeks. There is a lot of scatter because,although two days may have an equal total cumulative irradiation, the cloud patterns and direct diffuse ratios vary all the time, which is why the cumulative daily pumped water output and the overall daily system efficiency are not necessarily the same for a given cumulative energy input. High daily outputs for a given irradiation level imply more hours of sunlight at near the optimum irradiance level for the system, and vice-versa. Figures 33 and 34 show similar curves as for Fieures 3 1 and 32 respectively, but for the Arco Solar system in the Sudan. This is optimised for a higher lrradiance level (which is sensiblesince Sudan will normally have more direct sunshine than Mali) and this is confmed by the much lower level of scatter in Figure 34, showing that the solar regime varies much lessday by day in the Sudan over the period recorded, mainly due to the rarity of clouds there at that time of the year (the tests in Mali were conducted partially during the rainy season). 4.10 Power Conditioners and Maximum Power Point Trackers It is possible to install an electronic device between the array and the load or subsystem which will automatically optimise the load on the arrays What this device does is to effectively alter the load impedance to.match the optimum impedance of the array. It does this by &awing the voltage current ratio required by the load through the use of dc-dc voltage transformation; two systemstested (from Solar Electric International and from SOTEREM) incorporated power conditioners for this purpose. The SE1system usesa maximum power point tracker (MPPT). This monitors the voltage applied to the load at brief time intervals (and hence the power supplied) and usesa logic circuit to control a high-efficiency pulse-width modulated down converter (PWMDC) to match the array operating point to motor demand, (Re ference 28). The logic usesa ‘hill climbing” technique which causesthe system to hunt for the maximum power of the load condition, and then to hunt by a small amount (about 1%) either side of the optimum point. If irradiance levels,array temperatures, the pumping head, or any other conditions change, then the logic circuit will seek the new maximum power point. Significant benefits can be gained from a MPPT of this kind, but ironically, the most benefits would be for a badly designed system with an ill-matched array and subsystem, or one operating well off its design conditions. However, the MPPT also consumes a certain amount of power (about 4 to 7% according to Reference 28), and thereby represents an energy dram which must be set against the savingsit obtains. It also, at present, appears to be an expensive component, costing as much or more than the electric motor, so in terms of costeffectiveness,the cost has to be set against energy gamed. A good system with a well-matched centrifugal pump can be designed so that it naturally has a close match between the maximum power and efficiency locus of the array, but systemswith piston or other positive displacement pumps do not match at all naturaIly becausea positive displacement pump, at a given head, tends to draw an almost constant current (it is a constant torque device). Hence, the voltage needs to be changed to maxlmise the power drawn, which does not happen Note: Army andsubsystemefficienciesunknownas arrayenergy output not recorded. FIGURE32 - DAILY OU’IPUTOF WMPESGUINARDSYSTEMIN MALI : : I FIGURE33 -PERFORMANCEOF ARC0 SOLARSYSTEMIN SUDANV IXRADIANCE 98 ? * -? gr 1 : ?. : i : -. a. a : >. .* g! . 5. Z’. $4. a FIGURE 34 - DAILY OU’IWT OF ARC0 SOLARSYSTEMIN SUDAN 99 naturally with PV arrays under varying irradiance levels.The SOTEREM system usesa form of MPPT to compensate for this and no doubt MPPTs will have a useful role in conjunction with positive displacement pumps. The use of electrical storage batteries, with a charging circuit to switch off the motor when the battery is discharged and to switch off the array if it is fully charged, offers another form of power conditioning by permitting the motor/pump unit to operate under almost constant and optimised I-V conditions. However, as outlined previously in section 4.6, batteries appear to have so many problems associatedwith their use under field conditions which apply to micro-irrigation, that it remains to be demonstrated whether they would, on balance, be beneficial in this application. 4.11 The Pmjeet Design Study Mathematierl Model To investigate the matching of system components as part of the Project design study described in detail in the accompanying Project Report, the Consultants developed a computer-based mathematical model of a PV solar pumping system with which different component performance characteristics could be simulated and their interactions investigated. An outline block diagram of the model is given in Figure 35. The model contains the sets of equations which describe the performance of each component, and solvesthem numerically to yield the unique set of H, Q, T, n, I and V values for a given irradiance level and array temperature. The motor and pump characteristics used for the model were derived directly from the data obtained from iaboratory tests of motors and pumps, smoothed by various normahsation processes to eliminate the scatter that results from practical experiment. Typical curves from the laboratory testing programme have been used, e.g. Figure 23 (motor) and Figures 24 to 29 (pumps of various kinds). The model was validated by making it simulate actual systemstested in the field, (the subsystemsof which were laboratory tested). Figure 36 illustrates a validation result comparing actual field results with tield simulation results (of the Arco Solar system tested in the Sudan). A sufficiently satisfactory correlation was obtained so that the general conclusions drawn from simulated runs of modified or different systems can be considered as reasonably accurate. The model has a solar input derived from typical daily recorded irradiance data, and recalculates H, Q, T, n, I and V for the specified system at predetermined intervals through the day. It also calculates the pumped output assumingquasisteady state conditions during the time interval used and adds the output to those from the previous time intervals to arrive at an integrated value for daily pumped output, as shown by the Total Volume curve of Figure 36. To allow a useful comparison between system options, two standard “solar days” were used as inputs: one was derived from Sudan data for a typical arid region, with a high proportion of direct radiation and virtually no passing clouds; the other, derived from Phikppine data had a higher level of diffuse with intermittent clouds in the afternoon. 100 PBFuwRMANCEPARAMETERs BASIC MODULES STATE VARIABLES TotaI and diffuse lrradiance. t SOLAR INPUT1 Climatic conditions Date and time Current and Voltage output Torque and Speed output I PV ARRAY v MOTOR Tracking Concentration Temperature Powe: conditioring Batteries Inverter Transmission I TP-tbihead Head and flow output Volume of water Cost per Unit Volume SpeciIlc Capital Cost SYSTEM ’ CHARACTERISTIC, 4 t COST OF WATER 1 Static head Pipework properties Capital cost of system components FIGURE 35 - BLOCK DIAGRAM FOR PHOTOVOLTAIC SYSTEM MODEL 4 IO II 12 I 13 I4 lima (hours) FKiURE 36 - VALIDATION OF PHOTOVOLTAK! SYSTEM MODEL (PUMP OUTPUT) t 12 To derivea true cost of unit output of water, it would havebeennecessaryto make a largenumber of assumptionsnot directly related to systemconcepts,suchasthe pattern of use,solar variation through the year, discount rates,economiclifetimes, etc. To avoid these assumptionsa simplification was introduced in which the capital or fit cost of the systemwas related to its expecteddally output under the standardsolar day in order to expressthe investmentin equipment of a particular type required to produce a given number of kilo-joules of hydraulic pumped energyper standardsolar day. This givesa relativemeasureof output costs from different systems“ail other things not related to the systembeing equal” i.e. a measureof their cost-effectiveness. The parameterused was called “Specific Capital Cost”* and was computed in units of US$/kJ per day. The SpecificCapital Cost is a reasonablemeasureof system“goodness”sincefirst cost is in the end the dominant cost for all solar pumping systemsand it can fairly be assumedthat recurrent costsare independentof the concept used. The cost data used in the model were preparedon the basisof the cost of componentsin a typical developingcountry for large volume orders. Arbitary factors which enter into the price of particular manufacturerswere eliminated by referenceto the power, weight and sizeof the component concerned. A programmeof model runs was conducted,simulating many of the systemstested in the tield and then simulating the effects of variousmodifications on a typical system.The kinds of modifications investigatedwere:0 static headvariation, to find the optimum operating headfor eachsystem 0 solarday variation, to fmd if somesystemsWCI~ rather than direct sunlight conditions or vice-versa. 0 reoptimisation of the PV array to fmd the best series-parallelmix of cells to minimise the Specific Capital Cost and also, to maximise the output with the original power rating. The conclusionsin most casescould only be implemented through the use of moduleswhich are generallyavailable becauseof the non standard number of cells per module required. However,if the solar pumping systemwere to be mass-produced,then the production of specialmodulesmay be justified. 0 impedance matching, to investigate the use of electronic Impedance matching deviceswhich automatically reoptimise the subsystemI-V requirementto match the maximum power locus of the array by in effect acting as a dc-dc transformer. Such devices,however, cost money and absorba proportion of the power. Therefore the model wasusedto investigate their cost-effectivenessunder different cost and efficiency assumptions. 0 the effect of sun tracking, either cantlnuously or intermittently (manually) was investigated.Continuous trackir!g is obviously best in terms of maxhnisingthe output, but it is much more expensive.Therefore the model evaluatedthe cost-effectivenessof the different options. * SeeAppendi 2, pageA8 for defmition of SpeciticCapital Cost. ’ hnisedfor hazy 103 0 &tally, the optimum pipe diameterswere investigatedby trading off the reducedresistanceof larger diameter pipe againstits extra cost, for a given length of pipe with different systems. Theserepresentjust a few of the areasthat can usefully be investigatedwith a model of this kind. At the time of preparation of this Report the field and laboratory data haveonly beenavailablefor a short period and therefore the full potential of the model remainsto be exploited in future Phasesof the Project. 4.12 Resulti obtained from Model Testing The Project Report (chapter 4.4) discussesthe resultsobtained in detail so it is only intended here to give an outline description of them in so far asthey relate to the generaldesignof solar PV pumping systems. The Arco Solar system was used extensively as a subject for the model investigations,becauseit is a simple and conventionaldesignwhich appearedto be reasonablewell but not perfectly optimised becauseit had a better than average, but not the best, efficiency; and becauseit seemedto be potentially one of the more cost-effectivesystems.A well documentedand comprehensivedata basewas also availablelargely due to the clear sunny conditions that prevail in the Khartoum areawhere the teststook place.Therefore the examplesof what might be possiblein the way of improvementsusethe Arco Solar systemasan example, but could equally apply to virtually all the other systemstested. Figure 37 showshow the efficiency of the systemvarieswith head; clearly the system as supplied was effectively optimised for about 7.5m, and performed reasonablywell for headsin the 6 to IOm, range.The array voltage was then “altered” in the model by increasingand decreasingthe cell string length by units of IO%,keeping the actual number of cells and hencethe power rating constant. To do this in practice would probably neednon-standardmodules,which would explain why the systemswere often not perfectly optimised. A 13%improvement in maximum efficiency wasobtained with the model of the Arco Solar systemby reducingthe array nominal voltageby 20%.A secondopthnisation exercisewas to changethe array power (number of cells), keeping to the optimum voltage, to obtain the most cost-effectivesystem;i.e. trading off array cost againstoutput - a smallerarray might meana reducedoutput, but the cost might alsogo down by a proportionately greateramount. With the presentday array cost assumptionsbuilt into the model, the Arco Solar system actually became more cost-effective accordingto the model with one lessmodule in the array. Clearly asarray costs change,the optimum power rating for cost-effectiveness will alsotend to change. Figure 38 shows how the cumulative dally output varieswith head using the “Standard hazy day” and the “Standard clear day” solar input. If the system efficiency remainedconstant at its maximum level, then a hyperbolically increasing output curvewould be expectedasshown by a broken line in the Figure. But due to declining efficiency at lower heads,the output tends to be almost linearly related to head;this kind of effect will of courseapply to most other systemstoo. ,,’ Supplirr: Standard l fficirncy Arca Solar claor day (*A) With optimum voltape Efficirncy on qrorr 0 2.5 5.0 is basrd cell area. 7.5 IO.0 Static head (ml iTGURE37-VARLATIONOFDAILYOVRRALLSYSTRMEFFICIENCYWITRHEADFORARRAY OPTMLSEDSYSTEM I Oally pumpad volunu (m3) System : A Sumliu : Arco S&r I I I I I \ \ \ \ \ ‘Y Outcut at constant rfficiancy \ \ \ \ x \\ / Standard hazy day-/ F \ 2.5 FlaJRE 38- VARlATlON 50 Head (m) OF OU’ll’UT WITH READ FOR PHOTOVOLTAIC Standard char dey 106 The useof a MPPT can partially compensate,but not to a great extent. This is becausethe pump itself variesin efticiency at different heads,(seefor example Figure 24), and this affects the system efficiency more profoundly than any impedancemismatch. The effect of applying a MPPT to the Arco Solar systemis shown in Figure 39. A MPPTwith 10%parasiticpower drain is ahnostequal to simple atray voltageand power optimisation (which of coursecostsmuch lessthan the MPPT) although a hypothetical cost-freeMPPT does improve the system maximum efficiency, but mainly closeto the designhead. One of the most cost-effectiveways of improving the daily output (and reducing SpecificCapital Cost), wasoccasionalmanualtracking. This wasdiscussedin detail earlier in section 2.7.3 under the headingof “system efficiency”. An initial appraisalwasmadeof the effect of vailation in pipework diameter on system economics.The result is shown in Figure 40 and confirms the dramatic importance of not usingpipeswhich are too small. Purchasersmight be tempted to savea little on the capital cost by acceptingsmall pipes, but this would be totally vitiated by the cost of the energy consumedin pipe friction. The SpecfticCapital Cost is sensitivebecausepipe friction is inverselyproportional to the fti power of diameter whereaspipe costsare related to the squareof diameter. Footvalves,sharpbends,kinks, valvesand unrecoveredexit lossesall add to the lossof energy,and attention to detailed designis required if theselossesare to be minim&d. Table 8 summarisesthe main fmdings in connectionwith the investigationsof the modelledArco Solar system.It can be seenthat voltageand power opthnisation can obtain a 12%reduction in SpecificCapital Cost. Manual tracking of the sun (two adjustmentsper day) reduces the SCCby 6%. But the combination of the abovetwo improvementsreducesSCCby asmuch as32%.The useof a MPPT with manual tracking (with 10%losses)increasesthe daily output from 32% to 39%but is lesseffective in terms of reducingthe SCC,becauseof the quite high coat of the MPPT. It must be stressedthat these results are indica’ive of the relative merits of different options asrevealedby the model; they apply only under “Standard solar day” conditions. It remainsto be testedunder the next Phaseof the Project, how well the improvementspredicted by the model can be achievedin practice, by making the modifications that appear to be beneficial to actual systemsand reteatingthem. 107 Swplirr : Arco Solar Standard clew day System with MPPT. 3.5 i-1 Overall system efficiency et%:, 3.0 I - 2.3i- 2.0 I.5 <=- 0. 0 2.5 5.0 I 10.0 I 7.5 Static FIGURE 39 - VARIATION OF DAILY OVERALL SYSTEM EFFICIENCY WlTH MAXIMUM POWER POINTTRACRER head (rnj WITE READ FOR SYSTEM I Static brad = Sm standard, char day *ific Capital Cast. (#I LJ perday! 0 2i is i0 Pipe diomrtrr (mm) FIGURE 40 - EFFECT OF PIPEWORK CHANGES ON SPECIFIC CAPITAL COST Id0 109 EffectlveU ) SpecSc MODIFICATION capital efficiency 6) None - Systemas supplied Voltage and power optimized by changing array cells series/parallel arrangement MITT (10% losses) (maximum power point tracker) Manual ;trackingof sun (2 adjustments of array position per&Y) Voltage and power optimized plus ma”...., rmrlrin~ MPPTplus manual sun tracking Notes: Cost (SFJ per day) Daily Output Sm head (standardclear &Y) (m3) at Effect compared with basic system, on Vohune Spec.Capital pl:mped cost 2.2 3.4 44.9 1.oo I .oo 2.6 3.0 46.6 (2) 0.88 2.5 3.1 51.2 1.14 0.9 1 2.9 2.6 57.1 1.30 0.94 3.3 2.3 60.3 (2) 0.68 3.1 2.5 62.3 1.39 0.74 ( 1) basedon irradiati .‘n m fixed array (referencedto cell area). (2) power is reducer:,&:ld so pumped volume is not comparable with basicsystem TABLE 8 - RESULTSOF MAKING ‘IMPROVEMENTS’TO A PV PUMPINGSYSTEM BY USMC TEE MATHEMATICAL SIMULATION MODEL :,::,: 110 5. SOLAR PUMPING TECHNOLOGY - THERMAL SYSTEMS 5.1 Isstory Solar thermal power systemsare by no meansa new development;a workable solar steamengineof severalhorsepower,heatedfrom an axicon focusingcollector, was developedby Mouchot and Pifre in Franceduring the 1870’s.Their enginewas demonstrateddriving a small printing pressat the ParisExhibition of 1878. Willsie and Boyle built a number of solar steamengineswith outputs in the 6 to 2Ohprange in the USA between 1902 and 1908, while their contemporary, Shuman,deveIopedthe fmt steam engine to be heated from a flat plate solar collector, also in the USA. In 1913 Shumanand Boys installed a SOhpsolarpowered irrigation pumping engine near Cairo in Egypt, using linear parabolic collectors.This pioneeringsolar irrigation system(and other contemporarysystems in the USA) functioned reasonablyeffectively until after a few yearsthey were all supersededby the introduction of internal combustion enginesdriving pumps and/or by mains electricity supplies,which at that time were becomingcheaperand more convenient,(see,for example,Reference30 for a more detailed historical account). The early solar-poweredsystemsmentioned abovegenerallyusedsteamasa working fluid; therefore concentrating solar collectors with modest concentration factors were usually usedin an effort to achievea reasonablecompromisebetween attaining a suffcient steam temperature for adequateengine thermodynamic effciency yet retaining a low enough concentration factor to mlnimise the accuracyneededfor solar tracking In contrast, the majority of modem small solar thermal poweredsystemshaveused heavy organicvtpours of the kinds commonly usedin refrigeration and air-conditioning systemsas working fluids. Theseboil at a temperature low enough to permit the useof non-tracking, flat plate solar collectors at the expenseof greater complexity in the designof the engineexpanderand rather low thermodynamic efficiency and consequentlya larger collector arearequirement to achievea given output. There are difficult technical decisionsto be made when defining the designphilosophy for solar thermal systemsdue to the conflicting requirementsfor achievinggood collector efficiency and good engineefficiency. Experimental work on solar thermal enginescontinued in many countries,but it was not until the 1960’s when it becameclear that solar power might have widespreadapplication in arid, sparselypopulated regions,particularly for water pumping, that seriousattempts to developviable systemswere againmade,mainly in Franceand Israel. At the presentstageof developmentthere hasbeenno clear demonstrationthat there is any single optinum approachto solar thermal powered systemdesign.In any case,factors suchasthe sire of the system(power output), application and manufacturing environment, must inevitably introduce important constraints. Despite the long history of solar thermal powered systems,there are still no manufacturersproducing systemson a commercialbasisin the world market. In the Consultantsopinion, none of the systemscurrently availableis asyet sufficiently developedto be viable in comparison with many other energy conversion LOCATlON POWER (KW COLLECTOR (ma ) DIRE Mali TABALAK Niger DELIVERY FLOW RATE (ml P) WATER HEAD (ml DAILY OPERATING TIME h IO 3200 1800 a 6-11 5 400 150 6 5-6 300 6 5-6 KARMA Niger 10 BAKEL Senegal(l) 32 1870 600 10 10 25 (1) 1500 150 54 5-6 SAN LUIS DE LA PAZ Note: (1) In association with Thermoelectron AREA IRRIGATED (W Corp (USA) TABLE 9 - LARGE-SCALE SOFRJZTES SOLAR THERMAL PUMPING INSTALLATIONS APPROX COST (6) 150 I ,640,OOO 100 540,000 s. 35 r 1 ” E 3 3c ‘zi E ii g 25 Likely -Thermal 20 Practical Efficiency Range 5 15 t IO 5 f 50 too IS0 200 250 Heat Source Temperature lGURE41 300 (“C ) -COHPARlSONOF~ORETICALCARNOTEFFlCIENCYWlTHEFFlCIENCIESOBTAINED~PRA~ICE(TC=2~UC~ ,,,, ,,,~, ,, ,,, 113 systemsunder free-marketconditions. Neverthelessthere are a number of prote type devicesthat could possibly offer prospectsfor developmentinto a viable pumpinem~. 5.2 Existing !bbr TIKNlal Pumping In.9talIations A number of manufacturershavesmallscale solarthermal pumping systemsunder developmentand severalhaveprototype field installations under test. The French company, Sofretes,is the only manufacturerhoweverwith a significant number of actual field units installed. The units are rated at about 1kW (mechanicalpower to pump) and employ flat plate solar collectors and a reciprocating organic fluid Rankine cycle engine operating at about 80°C. Overall effciency is about l%, (solar energyto shaft power). Thesesystemsare elaborsteand well-engineered,but it is known that for variousr:asonsmany havefailed to operatesatisfactorily in the field. The capital cost of thesesystemsalso appearedto be too high for economic operation and it is understoodthat the majority was fmancedasFrench overseas aid. Pumpingsystemsof this sizeand type are no longer being manufacturedby Sofretes,asthe company hasdecidedto concentrateits solar thermal resourceson considerablylargersystems,of 5 to 1OOkWrating, and to use photovoltaicsfor smaU-s&esystems.Someof theselarger Sofretesinstallations are listed in Table 9. 53 Thamdi?fehcyofButEnginw The theoretical maximum efficiency for conversionof heat energyto mechanical work was fii defined by Camot as q,=-h-Tc Th where This the absolutetemperatureof the heat sourceand Tc is that for the heat sink. In practice, evenwith good design,it is not possibleto obtain an efficiency anything like asgood as the theoretical Camot efficiency nc and in most casesthe true thermal efficiency for convertingheat to work will be in the range0.3 to 0.6 of uc. Figure 4 1 illustrates this relationship for the temperaturerangeof relevance for solar thermal systems. To date there are three main types of thermal systemswhich haveshown potential for small-scalesolar pumping applications.Theseare: (a) low temperatureorganicvapour cycle devices(SO0- 15OOC); 0-O medium temperaturewater vapour (steam) cycle devices( 140°C 3OOoC);and (@ gascycle devices(Stirling cycle), using air, hydrogen or helium asworking fluid, requiring input temperaturesgenerallyin excessof 3OOOC. It should be noted that the vapour or Rankme cycle which applies to organic vapour and steamenginesis inherently lessefficient (i.e. a smaller percentageof Camot efficiency) than the Stirling gas cycle which the Stirling engine approximately follows. 114 In eachof thesetypes, the basicprinciple of establishinga flow of heat and extracting a proportion of this aswork will apply, asillustrated schematicallyin Figure 42. Normally the heat sink will be the water that is pumped from the ground (although in somecaseait may be the atmosphere)and the sink temperatureis unlikely to be lessthan about 2SoCin the countriesof main relevanceto solar pumping.The greaterthe temperaturedifference betweenthe heat sourceand the heat sink, the greaterwill be the thermal efficiency of the heat enginewith less solar energyrequired for a givenwork output. It would thus appearthat a solar thermal system should have as high an input temperature to the expanderas possiblein order to maxhnisethe temperaturedifference betweensourceand sink. Unfortunately, the efficiency of solar collectors decreasesas the temperature betweentheir output and ambient increases,becausecollector heat lossesincrease in proportion to this temperaturedifference. Therefore the optimum requirements for heat engine efficiency are opposite to those for solar collector efficiency. As a result of the conflicting requirementsof heat enginesand solar collectors, varioustechnical trade-offsare required which result in a number of combinations of different types of solar collectors and engines.There is no obvious,clear cut winning approachand this is the main reasonwhy there is such a variety of different solar thermal systemsunder development. 5.4 so&r TheNlal couectors Typical curvesshowing the variation of collector efficiency with temperature difference for varioustypes of solar thermal collector are givenin Figure 43. From thesecurves,it can be seenthat simple single-glazedflat plate collectorsbecome particularly inefficient if expected to produce heat at temperaturedifferences greaterthan about 750C. By adding sophisticationto a flat plate collector in the form of double glazing and special selectivesurfacefinishes that reducethe emissionof heat from the absorbersurface,acceptableefficiency with temperature differencesin the range80 to IOOOCbecomespossible.However, to obtain temperaturedifferencesin excessof 1OOoCwhile retaining acceptableefficiency demandsthe useof a concentratingsolar collector. A concentratingsolar collector must be alignedwith sufficient accuracyto ensure that the focus coincides with the heat absorber.The greater the degreeof concentration, the more accuratelymust the concentratorbe alignedand of course any suchdevicehasto follow the sun during the courseof the day to keep the focus coincident with the absorber.This introducesthe complication of mounting the concentratingcollector in such a way that it may be tracked and an accurate tracking mechanismis also needed.Severalmethodshavebeendemonstratedfor tracking, including electronically controlled electric servodrives(wherephoto-cells are usedto sensethe alignment with the sun); clockwork drivestimed to follow the sun correctly; and a number of promising devicesin which misalignmentis detected by a sensorcylinder containing a liquid/vapour combination which is normally just shadedby the concentrator mirror. If the sun illuminates the sensor cylinder, the heat boils the Uquid and pressurisesthe vapour which drivesa servo mechanismto move the mirror sufficiently to bring the sensingcylinder back into a shaded position. Some of these tracking methods may also he applied to concentratingand evento flat plate photovoltaic arrays,asdiscussedin Chapter 4. r-l Hoot Scurca (Sdor Cdloetul I Hml Enqinr Work . Tmmmission FIGURE 42 - SCHEMATICARBANGBMENl’ OF A THERMAL SYSTEM 116 Figure 43 showsa comparisonof measuredperformanceresultsfrom Reference3 1, and illustrates that vari& line focus collectors can achieveefficienciescf around 70% at temperaturesof up to 3000C. By comparison,a double glazedselective surfaceflat plate collector is rather inefficient. Major technical problemswith tracking concentratorslie in obtaining a sufficiently &able tracking mechanism,in mounting the reflector array in such a way that it tracks easily but is not influenced by strong winds, and in providing a mirror surfacewhich will withstand such effects asram, dust, bird excrement,ultra-violet radiation, thermal shock and rough handling. It is howeverimportant to note that tracking concentratorsdo haveone major advantageover fixed flat plate units In addition to obtaining a higher temperature In that they can receiveconsiderably more solar energyduring the courseof the day, particularly in early morning and in the evening,asthey turn to face the sun (asexplainedin section 2.7, seeFigures 4A and B, SA and B). The fvted flat plate collector surfacewill receiveattenuated solar energyearly and late in the day due to the acute angleof the sun’srays. Theoretically, a tracking collector will receive“/2 times the solar radiation per unit areaper day than a fixed one. In practice, an evengreatergain in output, sometimes exceedinga factor of two, hasbeen observed.This may be accountedfor by consideringthat, in addition to the geometric effects, fixed flat plate collectors havegreater reflection losseswhen receivingsolar radiation at an acute angle. tracking collector may thus be expectedto receivesignificantly more energyper day per unit areaof collector and be able to start up much earlier and keep going much toter in the day. If the difficulties of constructing a reliable and robust tracking concentrator can be overcome, there are substantial technical and economicbenefits to be gained.Considerationmay also be givento mounting flat plate non-concentratingcollectors on gymbals to permit approximate manual tracking; by turning the collector eastwardsin the morning and westwardsin the afternoon significant gainsmay be expected,(SeeFigures5A and B). A . Most concentrating solar thermal collectors availabletoday are of the parabolic trough type, achieving temperatures of between 100 and 3000C. Higher temperaturesare possibleby either using parabolicdish mirror collectors,which havea point focus rather than the line focusof the trough, or by usinga field of tracking mirrors individually aimed at a central “power tower” target. Although quite high engine efficiencies could be achievedfrom systemsof this kind, the tracking requirementsare more precisewhich makesthese approachesappear difficult for small thermal systemsrequired for this Project. Howeverone potentially useful application of concentratingcollectorsis the use of a parabolic dish collector with a Stirling engine system. At least one manufacturer(Sunpower Inc., USA) is developinga systemof this kind and it is possiblethat the gains offered by the use of a Stirling cycle heat engine wili compensatefor the more rigorous tracking requirements. For many applications,suchasindustrial processheat and power generation,concentrating collectors have been shown to be more cost-effectivethan flat plate collectors. Parabolic trough collectors havebeen used for relatively large scale water pumping systemsin the USA although only one systemat Coolidge,Arizona, hasperformed satisfactorily and is still in operation. Peoh dftciency % under noostime solar irmdimut 60 U.S.A long term qool for line focus collector 70 cmcen~rotor 60 caIIeclor. Resulls fmm Sondio Not&al Laboratories R*port 6AND 600665 April 1960 with (ypicol 50 General Atomic FMSC 40 Scientific Atlanta FFMC flat plale and evacuated 30 Typical evacuated tube collector tube collector ChamcDristics 20 super-imposed Typical rinple QlOZed selective surface collector Typical double glazed / seIec1ive surfacr flat plale collector / IO 50 100 I50 200 250 300 Fluid outpul lemperalure FIGURE 43 - COMPARlSON OF SOLAR COLLECTOR PERFORMANCES (‘C ,, 118 Some manufacturersand systemsanalystsconsider that the use of evacuated collectorsmay prove to be the most cost-effectivesolution for systemsrequiring moderately high temperature. They have been shown to have good collection efftciea*:y,derived from their low heat losses,and severalmanufacturersin various parts of the world havestarted limited production. No estimatesareasyet availabh of their probable manufacturing costsfor large-scaleproduction and for the present they must be regardedaspromisingbut unproven. In contrast, solar ponds havebeenproposedasa simple, cheapmethod of collecting and storing solar energy,potentially on a largescale.Such a pond is a shallow body of water, typically 1 to 2m deep,with a black bottom. Whenincident solar radiation penetrates the pond, some is absorbedby the liquid and a large pioportion reachesthe bottom and is absorbedby the black surfacewhich is consequentlyheated.Thus water at the bottom of the pond reachesa higher temperaturethan that nearerthe surface.Convectioncurrentswhich would normally developand equahsethis temperaturedifference are preventedby the presenceof a strong density gradient from bottom to top through the useof dissolvedsalts(or by plastic membranes).Many natural salt water pondsare known which exhibit theseproperties,and scientific researchinto their usebeganin the 1950’sin Israel. The principal aim of reaseachinto solar ponds hasbeenconcernedwith methods of extracting energyfrom the pond. Success hasbeenachievedby recyclingthe hot layers through a heat exchangerand temperaturesof 9OoChave been achieved. Currently a power generationsystemusing an Ormat turbine is operatingin Israel. In India solar ponds havebeen used for the production of salt and for biogas digesterheating. The’solar pond could be a useful collector for solar pumping systemsbut must be designedfor a specific location and is as yet inadequately developed.Work is currently in progressin the USA, funded by the Department of Energy (DOE) to establishthe feasibility of shallow solar pond-drivenirrigation pumping systems (Reference32). Presentindications are that for a large (12OkW)system, the shallow solar pond option is lessexpensivethan other DOE funded installations which useconcentratingcollectors. 5.5 Rankine Cycle Engines Nearly all solar thermal enginesavailableor under developmenttoday are based on the Rankine or similar cycle, in which the working fluid undergoesa phase changefrom iiquid to vapour hi a boiler, followed by expansionof the vapour through an expander mechanismto extract work, after which the vapour is condensedback to liquid and transferredby a feed pump back to the boiler to be reheatedfor another cycle. A typical Rankhte cycle systemin its simplest form is shown diagramma’icallyin Figure 44. The working fluid is directly evaporatedin the solar collector before passingto the expanderwhich drivesthe working fluid feed pump aswell asthe main water pump. The water is used to cool the condenser.The systemsbeing developedfor example,by Dormer (WestGermany)and Solar Pump Corporation (USA) utihse this cycle, with Freon asthe working fluid. 6 FIGURE 44 - SIMPLE RANKINE CYCIE 6 LEGEND 1. 2. 3. 4. Solar Collector Wonder (Enpine) Condenser Feed Pump 5. 6. Z a Water Pump Water Foot Valve Exchanger/ Circulofinq Evaporator Pump F(WRE 45 - RANlClNE CYCLE Wl’l’H INTERMEDL4TE HEAR EXCHANGER The systemshown in Figure 45 is sintilar except that water is usedin the solar collectors,with an intermediate heat exchangertransferring heat to the working fluid. This arrangementhasbeenadopted, for example,for the systemsdeveloped by Sofretes(France) and Ormat (Israel). Systemsoperating at temperaturesbelow about 15OoCgenerally use an organic working fluid, asthe useof water presentstechnical problems. Fluoro-carbonsof the Freon type, ascommonly usedin refrigeration and air-conditioning equipment, havethe advantageof lower boiling points at useful pressuresand higher vappur densities than steam, thereby permitting more compact systemswith higher efficiency. A few systemshavebeendevelopedor proposedwhich use organichydrosarbon vapoursas a working fluid (butane being a commonly applied one). The main objection to the use of this kind of working fluid arisesfrom their high flammability which could provea serioushazardin the event of a leak developing. Evenmore hazardousthan butane are working fluids such asn-pentane(proposed for one system),which hasboth a very low flash point and high volatility. The Freons (and other fluoro-carbons)on the other hand are non-flammable and reasonablysafewith regard to toxicity. Whenany organicworking fluid is used,it becomesesaentialto contain the fluid in the systemwith 100%reliability sinceeventhe smallestof leakswill eventually causethe vapour cycle to ceasefunctioning. Also, these fluids are relatively expensive,so it is worth minimising the quantity required to fill the system. Some systemshave been proposedwhich incorporate a heat store in the solar circuit, introduced betweenthe solar collector and the intermediate heat exchanger.The simplest form of heat store consistsof a large water or brine vessel, but somesystemsinvolve the transfer of latent rather than sensibleheat through the melting and solidifying of waxesor salt mixtures or the evaporationand condensationof a separatevapour. In this way a largequantity of heat can be stored and cycled through quite a small temperaturerange.It seemsthat this technology may in some casesbe an important adjunct to the use of small solar thermal systemsto provide a thermal buffer betweenthe rather variable temperatureof the output from the solar collectors and the fixed temperatureconditions which are desirablefor rurmlng a vapour expander.Naturally a heat storeinvolvesfurther complication and expense,and therefore requiresthorough evaluation before being incorporated into a small-scalepumping systemintended to meet the special requirementsof this Project. collector An interesting variation of the Rankine cycle systemis under developmentin India by Hindustan Brown Boveri in associationwith the Indian Institute of Technology, Kanpur, who claim that becauseof its mechanicalsimplicity, it is likely to be much leas expensive than conventional solar thermal systems. The most basic arrangementthey proposeis illustrated in Figure 46. A volatile organic fluid is used which is immiscible with water. The fluid is heatedto a temperaturein the flat plate collectors 1 such that it will flash into vapour in the flash tank 2. When sufticient pressurehas developedin the primary system, valve 3 opens and the vapour expandsinto chamber4 and displaceswater from the chambervia a nonreturn valve and rising main 5 to tank 6. Some of the water from tank 6 is circulated through heat exchangepipes in expansionchamber 4 and causesthe 121 4. --&j;ZAL l.7 L .*a -_- V 6. _-_---_F-. -_F-.- A.--------- - 4 t I ._--- LEGEND 1. 2. 3. 4. Flot Plate Collector Flash Tonk Volvr Expansion Chomber RUNDUSTAN - BROWN 5. Rising Main With Non Return Valve 6. Water Collection Tonk 7. Condensotc Return Tonk LIQUID PISTON RANKINE CYCLE SYSTEM .: ,,, : . . ‘: : .’ vapoufto condenseand collect on,the’surface,of the water (it is lessdensethan water and immiscible.with water in liquid form). This processdrawsanother supply of water from the’.weh via a non-return valve.The condensedorganic fluid is .recovered. when,&water level riseshigh enoughin 4, via a condensatereturn tank ‘. at 7, and returns under gravity to the primary circuit. In practice it appearsthat rather more complication thanthis is involved in achievingthe full cycle, sincethe ~, prototype includu a .number of further tanks and connections.Although the plumbing appearsquite complex, this device is attractive in that it has no mechanicalexpanderand pump, the vapour interacting directly with the water to be pumped, so it should be potentially reliable and long lasting providing the organic working fluid can be adequately contained. This is important since the prototype appearsto usen-pentaneasa working fluid and this is a dangerously llammable vapour. ‘_: . ..‘. ., ,,:122 . The Hindustan Brown Boveri systemis still at the prototype skate,but wassaid to be closeto being marketed by the manufacturerin 19?9. It is not be!ievedto have been launchedasa commercially availableproduct at the time of writing. Another type of low temperaturesimple deviceinvolvesthe evaporationor boiling of an organiclow boiling point fluid, which displacesmore liquid fluid to a higher level. The weight of the collected fluid is then usedto do work while it is returned to a lower level. Systemsof this kind are inherently largein relation to their power rating, but could ln theory be reasonably efficient. Two manufacturersare understoodto be developingdevicesof this kind, SunpowerSystems(USA), who offer the “Minto Wheel” and Grinakers(Pty) Ltd., (Johannesburg),who havea rocking devicecalled the “Camel”. The latter deviceappearspromising on the basis of the cost and performance figures quoted by the original inventor but unfortunately little or no information has beenreleasedby the licensees. The principle of the so-called“Camel” is believedto be asillustrated in Figure 47. The working fluid, Freon, is evaporatedin the flat plate collector (1). The vapour displacesliquid Freon from a reservoir(Z), which in turn travelsup a flexible pipe to the reservoirt3), where it condenses.Whensufficient condensatehascollected it overbalancesthe arm (4), thereby delivering somewater asthe arm is connectedto a piston pump. A valve,(S), then opensautomatically and returns the condensate by gravity to a lower reservoir(6), and thenceback to the solar collector (1). The consultantshave certain reservationsabout the inventor’s original performance claims for this device,but its simplicity, apparentrobustnessand low cost make it appearpotmtially attractive for use by farmers in developingcountries. The licenseewasunable or unpreparedto supply us with a systemto test to verify the performancethat had beenclaimed. Unlike photovoltaic systems,where there are not many options for using an electrical output to pump water, there are a greatvariety of options for using hot vapour or hot gasto drive an engineand pump. In fact many photovoltaic system designersuse “off-the-shelf’ commercially availablemotors and pumps, but a thermal system designergenerally has to designand develop at least the engine and often the pump and transmissiontoo, which probably accountsfor the less advancedlevel of development of smali-scalesolar thermal pumping systems, comparedwith solar PV systems. 123 LEGEND I. Flat Plate Collrctor 2. Liquid Displocrmsnt 3. Upper Reservoir 4. Boloncr 5. Valve Arm 6. Lower Reservoir 7. Water Pump FIGURJZ47 - “CAMJ5L”GRAVITY OPERATEDSYSTEM Rsssrvoir 124 AU Rankine or vapour cycle systemsrequire an expander and there are in effect three broad categoriesof device which can be used; small turbines, rotary positive displacement expanders and reciprocating positive displacement expanders. In addition to such desirable characteristics as long life and high reliability with minimal maintenance needs,an important expander characteristic is its expansion efftciency fisentmpic efficiency) which provides a measureof its effectivenessat converting heat into work. Unfortunately smail turbines tend to be inherently inefBcient as extremely high standards of surface fmish, coupled with very small clearancesand high rotational speeds,are necessaryif a good isentropic efficienc~~ is to be achieved.Therefore, positive displacement expanders are more common the sixe range of interest for this Project. Rotary positive displacement expandersinclude vane machines,screw expanders and analogousarrangementsto most common types of compressoror blower (e.g. Roots, Centric, Swashplate-axial-piston).Reciprocating devicesappear at present to ‘bethe cheapestand most reliable option at this scale.They can take the form of the traditional steam engine with a piston in a cylinder, such as the steam system bebsgdeveloped by Jyoti Limited fbtdia), or they can incorporate diaphragmsOT m&sock seals(generally preferred for organic vapottr working fluids where 100’;’ vapour retention is vital), or they can be liquid displacement devicessuch as the shown In Fiires 46 and 47. 5.6 StirBn~ cyck Eninal Stirling (or Eriksson) cycle en&es use a gassuch as air, hydrogen or helium as a working fluid and are generally capable of higher thermodynamic efficiencies than Rankine cycle enginesat a given upper operating temperature. Their heat exchange interfaces also can be smaller and therefore cheaper, but in practice Stirling engines require heat inputs at temperatures in excessof about 3OOoC,so that a solar powered Stirling engine would almost certainly require a high concentration factor point focussingsolar collector or power tower which in turn needsaccurate solar tracking. The Stirling engine was patented in 1816 by the Rev. Robert Stirring. John E&son patented an open cycle variant during the 1850’s. The Eriksson engine becamevery widely used for fractional horsepower applications (in preference to small steam engines) during the latter part of the nineteenth century, particularly in the USA, and was usually fuelled with wood or coal. An example using a parabolic mirror to focus the sun’s rays through a window in its absorber was reportedly demonstrated in 1872 and may have been the first solar powered engine ever.Widespreadelectrification rendered the small Eriksson enginesobsolete during the early decadesof this century, when fractional horsepower electric motors took over. In recent years considerable effort has been made to develop high technology Stirling engines capable of competing with diesel engines for automotive applications (advantagesof better efficiency coupled with less noise and less exhaust pollution havebeen claimed). Another development has been the evolution of the free-piston Stirling engine by Beaie and others. (References 33 and 34). At least two manufacturers are developing devicesbasedon the Stirling cycle which offer promise for small-scalesolar pumping applications. Sunpower Inc. (USA) are developing the “Beale enghte” and Metal Box (India) Limited in Calcutta are developing the “Fluidyne” pump originally invented by the UK Atomic Energy - - 125 Authority at Harwell. Both devicesare said by their developersto be close to commercialproduction; the Bealeenginewasin fact laboratory testedunder this project asexplainedbelow but the respectivedevelopmentschedulesof Metal Box (India) and the laboratory programmeof this Project madeit impossiblefor us to test a “Fluidyne” pump, although a reproduction basinwas seenworking at the manufacturer*spremisesin Calcutta, India. The “Fluidyne” pump in particular is worth describing in more detail. As illustrated schematicallyin Figure 48, water is causedto oscillate in a U-shapeddisplacer tube, one end of which is heatedand the other cooled. The movementof the water causesair trapped abovethe U-tube to be displacedthrough a closedair pipe joining the two sides;the air pipe acts asa regeneratorand storesheat from air asit is pushedfrom the hot side to the cooled sideand givesup its heat to the cool returning air. The alternate heating and cooling of a fixed massof air causes pressureand volume variationswhich not only excite the displacerwater motion in the correct phasing.but are alsousedto pump water asthe pressurealternatively drawswater from the inlet valveand expelswater through the delivery valve.The original “Fluidyne” pump asreported by Hanvell wasvery inefficient, but we are informed by the Metal Box Company that they have made a significant breakthroughand that an overall efficiency of around 3% hasbeenobtained (with electrical heat input under test conditions the overall efficiency wasabsar 7%). The presentpump is designedto be heatedby coal wood orges;clearly if reasonable efficiency can be obtained with a direct solar input, the extreme simplicity and lack of moving mechanicalcomponentswould make this devicevery attractive for useunder agricultural field conditions. Experimental work involving the use of the traditional mechanically-coupled. Stirling enginehasbeen reported by Puri (Reference35). Also, it is understood that Philips NV, in the Netherlands,havebeeninvolvedin the developmentof a small Stirling engine to power an electrical generator for running an irrigation pump, (References9 and 20), although it is believedthis programmeis no longer active. Another interesting type of Stirling cycle devicefor small-scalesolar pumping is the free-pistonStirling engine,or Bealeengine,under developmentby SunpowerInc. (USA). This devicewas testedunder the Project, and is illustrated in Figure 49. The thermodynamic working fluid is helium gaspressurisedto about S bar; low atomic weight gaseslike helium have particularly good thermal conductivity, especially when pressurised.The Figure illustrates the versiontestedwhich had an electrical heater:when solar-powered,the hot end of the enginewould be mounted at the focus of a point focusing parabolic collecior, (or it could be installed in a power tower). The principle of the free-pistonStirling engineis complex and dependson interactions betweenvibrating componentsconstrainedby springs.The stainlesssteel displacementpiston and the brassworking piston are interconnectedby springsand by the viscosity of the working fluid in such a way that they vibrate (with critical damping) 900 out of phasewith eachother. This shifts the working fluid (helium gas)from the hot spaceabovethe displacementpiston, through the annular wire matrix regenerator,(which takesheat out of the gas),io the cool spacebetween the displacementand working pistons.The gascools and reducesin pressureasa result, which drawsup the working piston and the connectedpump diaphragm. Heat in - - ---------__ -------- -----------_----- - Displacer _ ------ ---- Tube FlGURE48-PRINCIFLEOFFLUIDYNEPUMP ----- - 127 The displacerthen movesback downwardsthrough its natural vibrations and the working piston movesup and the working fluid is displacedback through the regenerator,where it picks up stored heat, to the hot space.The working fluid then reachesthe peak pressureof the cycle and drivesthe working piston and the pump diaphragmdownwardsand the cycle repeatsitself. 5.7 Tmmmkmionsand Pumps AR mechanicalexpanders(engines)will provide either a rotating shaft output or a reciprocatingshaft output. The former can be coupledto a rotary pump in much the sameway asdescribedfor photovoltaic pumping systemsusing a mechanical tmmmimion; hydraulic drive or direct drive being used where a suitable rotary pump to match the output speedis available.It is alsopossibleto drive an electrical generatorand transmit the output electrically to a suitableelectric pump unit, in which casethe commentson the relative efficiency of different electrical machines asgivenin section4.5. will alsoapply. Most of the commentsin section4.7 on pumps also apply with equal validity here too. Reciprocatingoutput expanders offer a convenient possibility of coupling directly to the pump, provided the frequency matchesthat required for efficient pump operation, but for the reasons givenin section4.7 suchsystemsare more likely to be appropriate for medium or high rather than low lift pumping. 5.8 Testing of Themtaf Systems Perhapsbecausethermal systemsare more complicated to designand assemblethan PV systems(with most, if not all, componentsrequiring to be purposedesigned) only one solar thermal small-scalepumping systemwasavailablefor purchaseon a commercialbasisand delivery within the timescaleof the first Phaseof the Project (comparedwith 10 different PV Systems). This sole thermal system,by Solar Pump Corporation of Las -legas,Nevada,USA, is illustrated schematicallyin Figure SO. It consistsof a simple, single-glazedflat-plate collector, (2), fixed to a support frame (I). The designworking fluid is Rl I (Freon I I) which is boiled in the collector absorber(3), the hot liquid Freon flashesinto vapour through a control valve (6) to the expander(7). The expander is located in the centre of the collector, which is a good designfeature to keep it warm; it consistsof a reciprocatingdiaphragmwith linked valvesand looks rather akin to the vacuum servounits usedto boost automobile or truck brakes.The expanderdirectly works a leverarm (8), which carriesthe pump rod connectedto the well pump cylinder assembly(12). The samelever arm (8) works a small feed pump which returns condensedFreon from the condenser(9) built into the support frame, to the solar collector. The condenseris cooled by water lifted from the well, which is delivered directly from the condenserwater outlet, (I 3). The whole unit is compactand well integrated, being 2.7m wide by 2.0m from front to back and 1.8m high. The Solar Pump Corporation systemwasinstalled near Khartoum in the Sudanfor testing, but due to delaysand problemsin commissioningthis system,(asmore fully explainedin the Project Report), it wasnot possibleto conduct more than superficial testing during the fit PhaselHowever,it demonstrateda StOiMss steal disatamnent oistm -weight 0~65kg. Annular w far gas transfer within cyliidsr wall. ElECtfkOl heater and thermal insulation. -I- I I- Wire matrix ragemratar in this ragIon in amular pap in cytii wall. LTfssms Water outlet. _ Piston ring. .// / Stainless steel cylinder. Water flow. Aluminium camactor to diamrogm. ,,rReMining v. r’o’ I thread to diiphragm. ring seal, / Citcllp rataimr. 6as filling vaive. Water inlet. FIGURE 49 - GENERAL ARRANGEMENTOF SUNPOWERINC. STIRLING ENGINE Support frame Collector frame outside Freon boiler Freon liquid distributor header Freon vopour header Vapaur pressure control valve Freon engine and valve Lever arm Freon liquid punp Pump operating lever arm Well pimp cylinder assembly Water discharge Arrows indicating water flow. Height : I .B metres Width : 2 ,7 metros Length : 2 0 m&es FIGURE 50 - SCHEMATICOF SOLAR PUMPCORPORATIONSOLAR PUMP 130 hydraulic pumped power output of 45W and an overall systemefficiency of 0.9%. The price of the unit wascompetitive with many of the PV systemstestedand the Cons&ants believethat the output and efficiency quoted aboveare not necessarily optimum and that there is somescopefor designimprovement.The systemwas alsogivenbrief tests at the manufacturer’sworks. The Pmject also gamedfmt-hand experiencewith solar thermal systemsby armngingfor teststo be carried out on prototype systemsat their manufacturers’ works at Ormat Turbines Ltd. (Israel) and Domier SystemGmbH (W Germany). In addition, a small Stirling engine pumping systemby Sunpower Inc., (USA) was tested for the Project at the University of Readingin the UK This work is describedin detail in the Project Report. The Ormat systemwasjust outside the scopeof this Project, havinga turbine using Freon asa working tluid, driving an eiectrical generatorwith a maximum rating of 4.7 kW. The engineand generatorefticiency was6%,implying a systemefficiency of about 2.2% if a collector efficiency of 60% is assumedwith a similar pump efficiency. The Domier systemhad a maximum hydraulic (pumped) power output of about 230W.The engine/pumpefficiency wasmeasuredat 3.5%, implying a likely system efftciency of about 2.0%. The SungowerStirling enginepump unit (SeeFigure 49) failed to perform to its manufacturer’sspecificationand then suffereda component failure which damaged it and cut short its test pmgramme.It is believedthat it may havebeendamagedin ttansit, which would account for its worsethan expectedperformance,evenbefore the component failure. Under test at ReadingUniversity it only developeda hydraulic pumped output of 23SW with an engine/pump efficiency of 1.8% (implying an overall systemefficiency with a point-focusingconcentratorof about 1.2%).The manufacturersclaimed a hydraulic output of about 8OW,which implies an engine/pumpefficiency of about 8% or an overall efficiency of about 5% with a point focusingsolar concentrator;this claim appearsplausible for a deviceof this kind operatingat temperaturesof between500 and 6OOOC. However,this free-pistonStirling engineprototype is a pieceof precisionengineer ing which is vulnerable to damagedue to the close fits and quite thin-walled componentscontained in it. Nevertheless,once fully developed,it may well be a compactand efficient small-scalesolar thermal pumping system. 5.9 Thenml SystemDe&r Studies 5.9.1 Introduction and Scope The Project thermal designstudiesinevitably were much more tentative in their approachand conclusionsthan were those for photovoltaic systems, becausethere has been very little experiencewith workable small-scale solar thermal systems, despite their long history. However, some interesting and perhapssignificant conclusionshavecome out of the study which is describedin the comparisonProject Report, and these are outlined below. 131 Most of the earliestwork on solar pumping wasconcernedwith the useof heat engines;indeed someof the fit achievementsin the useof solar energyin the early part of this century were with thermodynamic pumps. In spite of this previousexperienceit becamecleardining the early part of this Project that small solar thermodynamic water pumps havenot been developed to the stage where reliable systems can be obtained commercdy. It wasnot possibletherefore to obtain much fit hand experiencewith, or data from, working solar thermodynamic pumping systems.Data were obtained from only one complete system operatmg from solar power, that of the Solar Pump Corporation system,when tests were performed on the systemat the manufacturer’spremises.Accordingly it was agreedthat the Consultantswould undertake the necessarydesign studiesin order to: 0) determine whether it is feasiblethat solar thermodynamic pumps could competewith photovoltaic systems (ii) identify directions in which technical developmentscan be made taking particular account of what could be manufacturedwithin develop ing countries rather than what could be achievedwith advanced technologies.It was further agreedthat a mathematicalmodel should be constructedand usedto achievetheseobjectives. 5.9.2 The Model (0 Approach The operating characteristicsof solar collectors and heat rejection systemsare very sensitiveto temperature,and will vary dependingon the way the componentsinteract. This meansthat successfuloperation of a solar thermodynamic pump is critically dependenton closematching of the components. Comprehensivemathematicalmodelling of the performanceof a solar thermodynamic pumping systemwasbeyond the scopeof this study. Indeed this could not be possiblebecauseinsufficient data are availableto validate sucha model. It was thought, however,that it would be feasibleto usea simple model to comparethe likely performanceand costswhich might be achievedfrom different combinationsof solar collectorsand engines. The limited validity of this approachis clearly recognised;in particular the dynamic behaviourof thermal systemsis of great importance but resourceswere not available to study this. TABLE 10 - SOLAR COLLECTOR TYPES USED IN THERMAL DESIGN STUDIES 134 System Option Flat plate solar collector Freon omanic Rankine cycle engine Non-tracking Zkwh/day en&e sohr collector output Area required : 36.3m2 Cost basis : $108/m’ Component cost EIigiltC $3920 Equivalent swept volume : 1.11 litres Cost basis : 8 525/litre Component cost $ SC0 Additional $ 690 futed costs Equivalent tube length : 50.7m Cost basis : s7/m Component cost $ 355 Pump (Khn depth) Fixed cost $ 400 WeIl haad assembly and piping Fixed cost $ 200 Total cost TABLE 1 1 - EXAMPLE OF THERMAL SYSTEM COSTING. (1981) 8 6145 135 5.9.3 Function of the Thermal SystemsModel The model takes the performancecharacteristicsof known solar collectors and thermal enginetypes and combinesthem. Figure 41 showshow the ericciencyof enginesimproveswith increasingtemperature,while Figure 43 indicateshow thermal collector efftcienciesdecline with temperature. When collectors are combined with engines,an optimum operating temperatureexists,where maximum systemefficiency is obtained, which representsthe best trade-off between the contIicting requirements.Figure 51 indicates the effects on the efficiency of Rl 1 (Freon 11) Rankine cyde systemsusing a) a single-glazedsimple flatplate solar collector and b) usinga line-focusingmoderateconcentration factor collector This latter exampleis about the upper temperaturelimit at which R 11 can be used. Referenceto Figure 6 in section 2.7 givesa useful indication of where the principal lc,asesoccur on a complete small-scalesolar thermal pumping system. It must be rememberedthat the overall efficiency figures of Fii 5 I refer only to the collector and Raukins eugineprbr? mere:; a water pump plus the transmissionand pipework associatedwith it will, as shown in Figure 7, introduce further lossesso that the “overall efficiency” of Figure S1 needsto be reducedby from 40 to 80% for a complete smaU-scale pumping system. For comparison,Rankine water-based(steam)systemsusing the same kind of Line-focuscollector are also indicated. In one casethe steamis condensedat atmosphericpressure,(i.e. 100% as in an open cycle systemexhaustingto atmosphere)and in. the other case,sub-atmospheric pressurecondensingis assumed,at 55oC. Lower temperaturecondensinghas the sameeffect ashigher temperature boiling in widening the temperaturedifference between the heat source and sink and thereby improvesthe thermal efficiency. Both water based options would havean optimum operatingtemperatureabove200°C, but for the purposesof the modelling exerciseit wasassumedthat this is the upper temperature limit for reasonablysimple line-focus collectors and tracking systemsas weU as in relation to the pressuresand conditions that would apply in the engine.This is bearingin mind that very small systemsfor use by farmersin developingcountriesare being considered. The comparatively flat curvesfor concentratinglinear focus collectorsin Figure 43 might appearto suggestthey are radically superior to flat plate coUectors,but it must be rememberedthat the main casefor the useof tIat plate collectors is on account of their relativesimplicity and therefore aUegedlow costscomparedwith tracking systems. 136 It is also worth noting that so far asconcentrating,tracking systemsare concerned,their cost and complexity increaseconsiderablyasa function of the concentration factor (i.e. operatingtemperature) due to the need for more precisefocusingand tracking and to greaterdifficulties inherent in avoiding excessiveheat losseswhen transmitting the working fluid from the absorbertube(s) to the engine. Therefore the relatively high efficiencies obtainable from certain linear focus tracking systemsat temperaturesashigh as250 to 300°C may only be bought at a considerablecost both in terms of complexity and actual collector cost; however,with clever designit may also be possibleto achievesuch temperaturesefficiently, reliably and at competitive costs. The modelling exerciseattempted to introduce various cost assumptions in assessing the relativemerits of different systems.This is vital, sincea purely technical assessmentwill inevitably simply prove that higher temperature systemsare both more efficient and (consequently) much smaller;it would not say whether the greaterefficiency and reducedsize results in greater or lessercost-effectivenessor how the apparent costeffectivenessappearsto compare with small-scalesolar photovoltaic pumping systems. The resultsof the modelling exercisewere presentedin terms of Specific Capital Cost, in exactly the sameway asfor the PV systemcost analysis;a valueof this is to allow direct comparisonwith solar PV pumping system LUJWds uncusseu III the pteceedingcnapter. S?ecitic Capital Cost is a measureof the investment required to install a system capable of producing a givenhydraulic pumped energyoutput under a standarddaily irradiation regime. The input/output requirement usedasa basisfor comparinga variety of collector and enginecombinations was: 0 0 solar input 6kWh/m’ per day (samepattern as for the PV model analysisdescribedearlier). engineoutput, with the aboveinput required to be 2kWh of shaft energy per day (which would represent 600Wh of pumped output per day if a 30% efficient pump were used.This is typical of the low headefficiency that would haveto be expectedusing a positive displacementpump best suited for use with a Rankine engine;at higher headsthe pump efticiency might reach60% or more, giving 1200Wh/dayof output from the samesizesystem). Sincet’lermal systemsare more likely to improve in efficiency if their size is increased,an analysiswas also initiated into the likely costsof a system designedto deliver IOkWh/day of shaft power under similar input conditions. There were many assumptionsthat necessarilyhad to be madeasto the effects of economiesof scaleon the systemcomponents,many relating to the pump, rather than the engineitself. The result is the product of multiplying theseassumptionstogether and in many ways Indicated little more 2bw RNmNE DNLV oumn BNBNE’tWE OFrNlJM aPERATNfi TIwBRAllJRE Y%’ Or~udcBmkheQcls(RIl) oqmkRdbleCyck(RII) [email protected]&eCyck(Rll) OTSmkRmkheCyck(RII) [email protected] Rr*he Cyck (RI I) Oqmb Rmldne Cysk (RI I) O~~mkRmklmCycla(RlI) olgmb lwbbw Cyck (RI I) Or~mkRuddneCysb(RII) [email protected](RI1) OrpnlcRmklM Cyek (RII) Stuns. .xden,u 60.16 bar Stmm, ccedmwr ‘3 I bar orllnlfFMdncCycb(RII) Slmm, cmdmwr eO.16 ba Stem. comknwr @ I ba Orpn*RmkhCyck(RII) Steam,mndemereO.I6b~r Sa.m, conde,,ur et bar OS& Rankhe Cycb (RI I) Sam. sondenvr B 0.16 br Sbmn. cmdmer 19 I bu slklllycycb Ckpdc Rhlnc Cycle (RI I) Skmn. Eondnurr 60.16 ba Slm. condmnwr 8 I br 3 4 5 6 7 I 9 IO II I2 13 14 is 16 17 1s 19 23 24 23 6tbkh6‘3Cb On~nnk RmldneCycb(R11) Steam. condenser QO.16 br Stem. condmssr @ I bar sudbl~cyck TABLE I2 - RESULTS OF THERMAL PEAX ENGINE mwnnoumrr coi,uNxoR AREA WA119 IO su 90 90 130 I30 I30 10s 106 l4l I30 I96 I99 IXI I96 I90 IS0 IS6 iii I96 I90 SW 150 I99 I96 500 I50 I96 I90 MO SYSTEM MATHEMATICAL (NOTB 2) CAPITAL CaBr SyEClFIC CAPl7AL COBT L 270 4991 274 .74 236 409 261 777 492 441 3S4 401 461 384 um 400 451 330 530 20.6 36.3 1S.S 32.5 S.I 12.9 10.3 3S.0 20.9 13.1 10.0 IO.8 lb.4 9.9 10.6 16.1 13.2 17‘0 23,s 7,s 8.3 12.6 4.4 8.6 310 382 362 u3 361 366 366 406 366 317 377 423 as 13.3 4.3 7.7 6.1 12.3 4.0 MODELLING 44410 6150 4340 6300 3680 4330 3210 t:70 43M 3490 3310 3850 4110 3290 3820 4070 3SM 4930 5520 3370 3860 4170 2420 3770 4180 4350 2660 2790 3230 3220 2060 2.9 I.7 2.1 I.5 2.9 2.0 I.6 I.3 I.0 1.9 1.5 1,s I.9 I.8 2.3 2.6 I.6 1.8 1.9 I.1 1.7 I9 2.1 1.2 I.3 I.5 I.5 1.0 138 than a larger systemwould be somewhatmore costeffective. How much more cost-effective cauld only reliably be determined through considerablefurther data collection and analysis.Sinceit is felt that the main value of this analysisis to show the relative merits of different system concepts in very broad and basicterms, and considerationof a largersystemdid not alter the apparentrelativemerits of the different conceptsove: the scalerangechosen,it wasdecidednot to proceedwith work on the lOkWh/day systemanalysis. For eachsystem concept considered,the model calculates: 5.9.4 0 the optimum operating temperature, 0 the areaof collector and sizeof engineand condensernecessary to achievethe required output with the input energy level specified, 0 and, hence the Specific Capital Cost to be expected for each system concept, (derived from cost data used in the manner shown in Table 13). Resultsof Thermal SystemsAnalysis The resultsare presentedin Table 12 and Figures52 and 53. Figure 52 givesan indication of the relative cost-effectiveness of the different system conceptsanalysed,so it involvesboth technical and cost assumptions. Figure 53, on the other hand, simply showsthe relative collector areas required for eachconceptand is therefore purely basedon technical considerations;collector costsare dominant in generaland therefore the area of collector neededis of interest. The greatestuncertainty when going from the assumptionsof Figure 53 to those of Figure 52 is the price differential betweensophisticatedpoint focus collectors and power towers and the lesssophisticatedline focus collectors (and their tracking mechanisms)and flat plate collectors.There is uncertainty on how mass-productionmight reduce the relative unit costs of the more sophisticated collectors compared with flat plate collectors; the analysisis basedon presentday costsand projectionsand comparestechnologiesat different stagesof maturity and scale of production. On the cost ‘wsumptionsused, which it is believedare reasonablycloseto presentday reality, it seemsthat the improved efficiency resulting from a greater temperature difference (either through a hotter collector or a cooler condenser)is cost-effective.In other words, the higher temperature and more efficient systemsinvariably show significant reductions in Specific Capital Costs compared with conventional simple low temperatureflat-plate collector systems. it should be noted that if the collector termiology usedin Table 12is unfamiliar, the types of collector referedto are specifiedin more detail in Table 10. ovrraii l ffftiency Coffrcf~ 70 m, rfflcicrcy Steam l n3inr,temp of condeneer = 55’C \ Lin4 f ocurln6 Coliecior efficiency \ 14 13 -- \ ---.. 60 \ * -- , /- 12 ii 50 40 IO 9 Single 6ie*-* I..” flu-alal . ,.-h call --.:ecfw efficiency Oieraii l ffici*ncy Strom mginr,trmp of condeneer = 100.C-Y \ ,6 7 6 Jo E ,5 /. Ourroii efficiency All l n6lne with elfqf. QiOred fioiplot. colircfor 20 .4 .3 .2 ha .I C I , I I I I I I I SO 60 70 60 SO 100 110 120 130 I i I I I I 140 IS0 160 170 I60 190 Coilrctor opefoffng femperoture CC 1 FlGfJRf? 51 - RXAMPLES OF SOLAR COLLECTOR AND ENGINE EFFICIENCY (Excluding pump, meckmicef tnmmiesion end piwwork loses) ,O 30 Ov*rdi elficlency (?a SPEClFIC CAPITAL COST @kJdoy-’ k( x4x6 X2 LEGEND 2-t The numbers rofrr lo the descriptions @IO X6 03 .I .s PC X 026 .o A Nan - Irochinq sYsfa+ toll olhers invoke lurching ) Siam as working fluid systemr use Freon II ) Stirling in Table 12. (all other Ranhine cycle cycle systems I.5 A 21 A 23 A 31 I-O fx 0 , ID0 IS0 2w FIGURE 52 - l?FFECT OF OFllMUM 250 OPERATING 300 TEMPERATURR I 350 400 Optimum 450 5’;OV Oprroling Tamperolure ON COSTS OF THERMAL SYSThS x2 X6 X4 Area Id LEGEND 3c The numbers @IS 2: X Non- frocking 0 Steam as worhinq fluid (all systems use Freon II ) A ,, Stirlinq cycle system (all in Table 12 others involve tracking ) other Ronkine cycle oyslems lS *I 2f refer to the descriptions 03 I! . .I? 10X6 922 26 - 30 IC I I t I OPTIMUM OPERATING L 300 200 100 FRXJRR 53 - RFFECTOF I TEMPERATURE ON COLLECTOR , I 400 Optimum AREAS OF THERMAL I I 5oo*c Operating SYSTEMS temp. . . ;. ,Certain factors stand out from the analysisand are highlighted by Figure 52:. . . a) . futed.tIat-plate collectors driving Rl 1 enginesappearto be about the leastcost-effectiveoption. This conclusionis at variancewith the trends of recent years as aspiring manufacturershave used preciselythis apparently wrong approach. b) tracking introduces significant economiesevenwith flat plate collectors, for the kind of reasonsdetailed in section 2.7 and illustrated in Figures 4A and B,,5A and B. Compareexamples2,4 and 8 in Figure 52 with their tracking equivalents1,3 and 9. cl systemsusing RI 1 asa working fluid appearcheaperthan similar systems(using the sametype of collector) with steam as a working fluid, evenwhen operatingat a lower temperature.This is mainly becauseRl 1 hasbetter thermodynamic characteristics than steam ~. and its useallows a ._at the temperaturesconsidered mucn smallersystem,asis also clearly evident from comparing areasof collectors of similar conceptsin Figure 53. However,the steam engine might be able to overcomeits size disadvantage through simpler and lower cost construction. d) Piston Stirling enginesoperatingat high temperaturesappearvery attractive, but they are still far from practical use and many “ifs” are built into the cost assumptions.The main reasonfor their apparentlow costsis the small size that resultsfrom their high efficiency. They are almost certainly the most efficient method of converting solar energy into shaft power currently available,and could becomecheapthrough massproduction if a successfuland reliable designis perfected, since the material requirementsfor them are much smaller than for other types of system.Against thesevirtues must be balancedthe cost and complexity of both the high temperature engine and of the high concentration factor collector with a very precise tracking mechanism(two axis). e) the Fluidyne liquid piston engine/pumpwhich operateson the Stirling cycle in the form produced by Metal Box Ltd., has maderapid progresstowards commercialisation. D central receiversystems(power towers) are usually associated with large-scalesolar thermal systems and have received surprisingly little attention for small-scaleapplications.They appearattractive for this kind of application, but are not in production for such small-scaleusage. B) the fixed E-W trough collector, (with occasionalseasonal reorientation in the vertical plane), appearsunattractive, being little better than a futed flat plate collector. 143 h) a simple parabolic trough (or linear Fresnel lens) concentrator appearsattractive for usein the medium to short term asit offers appreciableadvantagesover flat plate collectors and yet is quite a well developedconcept with numerousexampleson the commercial market, (intended mainly for heating commercialbuildings or for processheat). 0 the analysisof the sameconceptsused to deliver lOkWh/day produced the samegeneralcost patterns, but the differences appearedto be more marked. j) the useof evacuatedtube collectors resultsin a marked reduction of sixe comparedwith conventionalfiat plate collectors;with a ZkWh/day non-trackingsystemthe reduction was from 36 down to 13 m* . Many uncertaintiessurroundprojections of the future coat of mass produced evacuatedtube collector and how competitive they would be for this application in cost terms. Also, the durability of evacuatedtubes under field conditions remainsto be demonstrated. k) quantity production of systemshavinga RI 1 (Freon) engineand linafocus parabolic trough collector could result in a system havinga SpecificCapital Cost of S 1.S/kJ per day (and an actual capital cost for a 2kWh shaft power per day output of around $3300,. It should be noted that the tracking/concentratingthermal systemsthat appearfrom the analysisto be so advantageouscomparedwith flat-plate systems,can only utillae the direct beam component of solar energy receivedwhile the latter can usediffuse radiation too. This wasallowed for in the analysis,although solar data from an arid region (Sudan)were used. In areaswith a high proportion of diffuse solar irradiation, the difference betweenflat plate and concentratingsystemswould be narrower,but this would not be through the flat-plate systemsbecomingmore cost-e.ffective, rather than the concentratingsystemswould be less. 5.9.5 Comparisonbetweenthermal and PV small scalepumping systems Table 8 indicatesthat givencertain improvements,plus manufacturein reasonablequantity, it should be possibleat presentday pricesto obtain a SpecificCapital Cost of around $2.3/kJ per day from PV systems. This compareswith the XC of around 8 S/kJ per day for imperfectly optimised presentday PV systemsmanufacturedin very small quantities. However,it is highly likely that by the time PV systemsare manufactured in somequantity (i.e. throusandsof units) the price of photocells will havefallen by 50% to around SS/W pk and, aspreviouslyexplained,it is probablethat within a few yearsthe price of PV cells will fall to nearer Sl orS2/Wpk. 144 A price of 65/W pk (50% reduction in PV cell costs)reducesthe XC of a well optimised PV system from 2.3 to about S 1.6 kJ per day. If the price of PV cells falls to S2/W pk. then the PV systemSCCwould correspondingly fall to about S l.Z/kJ per day. Further reductionsin PV cell costsdo not havea very significant effect on PV Specific Capital Cost sincethe balanceof systemcostsdominate. Probably a XC at present dollar valuesof around S I/ kl per day is the absolutelower limit that could be achievedfor small, low-head,PV systems. Therefore it does seempmbable that well-optimised PV systemswill eventually havea significant, but not an overwhelming,edgeover the best small thermal systemsfor low-head operation. Since PV systemsare already a more mature produce, it seemsunlikely that well optimised small thermal systems will become available before the expected reductions in PV cell costs are realised.The best SCCassumptionsfor small thermal systemsare far more speculativethan those used for PV systems,since they assumesuch requirementsas a reliable low cost tracking system,etc. However,the thermal systemscost analysisassumedoperation with a positive displacementpump, which at low headswould probably be no more than 30% efficient. At higher heads,where a positive displacementpump could be 7040% efficient, the SCC for small thermal systemsis quite likely to be lower than that for low-headPV systems.Further, PV systems for higher headswould be relatively more complicated, requiring positive displacementpumps with reduction gearing,plus power conditioning electronics to obtain good optimisation, or multi-stage turbine pumps driven by a long shaft from the surface,or a submersiblemultistage turbine pump (which would require an extremely reliable brushlessmotor to be practical). Theseextra complicationswould if anything substantially increasethe SCCfor higher headPV systemsascomparedwith low head ones. In other words, low headoperation seemsto favour PV systems,while higher headsseemto favour thermal (assumingsuitable thermal systems aredeveloped).However,high headoperation (over IOm) wasoutside the scopeof the first Phaseof this Project and further investigationwould be * neededto arriveat a reliable conclusion. So far analyseshave only considered“conventional” PV and thermal small-scalepumping system technologies,in the sensethat all systems considered could be put together with existing components. What cannot consideredor predicted is the possibility of new inventions in either field radically altering the picture. Both fields offer scopefor such possibilities, and many claims are made for unconventional solutions. Since the conventional technologies appear potentially capable of becomingcost-effectivefor small-scalepumping, although they are some way off at present,anything that offers any substantialimprovement in cost-effectivenesscould hardly avoid being immediately within striking distanceof offering economic operation. The next chapterconsiderssome of the unconventional technologieswhich just might overtake the approachesdescribedin this and the previouschapter, and somewhich may be of generalinterest but which seemunlikely contendersfor success in this field. 145 6. OTHER SOLAR PUMPINGSYSTEMOPTIONS 6.1 lWOllUCtlOIl The pe two sectionshavedescribedsolar photovoltaic and solar thermal pumping systems,which ‘areat presentthe only option: that havebeendemonstratedeither asccmmerc$alproductsor which aresufficiently developedto be closeto commerciaiisation.For completenessa number of other developmentsor potential methodsfor convertingsolar energyinto pumpedwater power are consideredbelow. Some of these techniquesdo not appearto have any future for smallscale solar pumping systems,but others may, with further development, proveto havezdvantagcsover the systemsavailableat present. 6.2 -limmo.electIic Generators Tbeaedevicesutilise the SeebeckEffect in which an e.m.f. is producedwhen the junctions of a circuit made of suitable dissimilar metals are maintained at different temperatures,a principle that hasbeen widely usedfor temperaturemeasurementsin certain types of thermocoupleand for working thermostaticswitches. At least two manufacturersproducecommercial thermo-electricgenerators(Global Thermoelectric Power SystemsLimited (Canada)and Teledyne Energy Systems Inc. (USA)), but both thesesystemsare intended for ultra-reliable, virtually zeromaintenancepower generationin remote areaswith heating provided by propane gasin cylinders;neither supplier offers solar poweredsystems. Like photovoltaic devices,the thermo-electricgeneratoris a solid state devicewith no moving parts and consequentlycan be expectedto havea long working life and to be very reliable. Tropaneheatedunits havebeenclaimed to be 4 to 6% effh5ent and to work without any maintenanceor attention for periodsof up to six yearsat a time and to havea useful life of many decades.Typical power outputs are in the 10 to 300Wrangewhich would be appropriatefor the end usebeing studied.Considerablework is believedto havebeencompleted in the USSRon thermalectric generationusingsolar heating and typical systemefficienciesof 1 to 4% havebeen reported, with considerableimprovement thought to be possible, (Reference36). Thermoelectric generatorsrequire high temperatureoperation (over SOOOC) and therefore most experimental examplestested with solar power aremounted at the focus of a parabolic dish concentratingcollector. As with high temperatureheat engines,they are therefore dependenton the useof reiiable and economichigh temperaturetracking concentratorsand require regulardirect sunshine. Their costsat presentdo not appearto make them competitive with photovoltaic systems. 146 6.3 Tlmnlonic Generators Theseinvolve heatinga suitable electronemitting cathodesurfacein a vacuumor low pressureionised gas;electronsare driven off and collectedby a cold anode placedvery closeto the cathode.The principle is much the sameasin a thermionic diode or valve(tube). Becausevery high temperaturesarerequired (over lOOOoC), there doesnot seemto be much potential for their use for small-scalesolar pumpb. 6.4 Brayton (Gas Turbine) SoIar ThennaI Systems This processinvolvesthe continuous compression,heatingand then expansionof a gas(as in a conventional gasturbine). It is mentioned for completenessbut appearsto have no potential for small-scalesolar pumping becauseof the high temperaturesrequired and the practical problemsassociatedwith small turbines. 6.5 Photochemkal systems Ordinary plant growth by photosynthesisis the prime exampleof a photochemical system which forms the basisof all life on earth and which also formed most of our fossil fuels. Natural biomassutilisation is outside the scopeof this project, but a synthetic photochemical processwhich showssomepromiseis the photoelectrolysisof water, in which sunlight is used,in the presenceof variouscatalysts, to.break water down into its constituent componentsof hydrogenand oxygen, which could then be usedto power a fuel cell, or an engine. It is too early to giveany kind of fii estimateof the future potential of the catalytic photo-electrolysisof water processand artificial photosynthesisis even further from practical realisation at present,but if and when it is successfulit could revolutionise the use of solar energy. 6.6 Improved Efficiency Photovoltaic Technology Apart from the developmentof Cadmium Sulphide and Gallium Arsenide cells ah&y referred to in Chapter 4, a number of other innovative developments are in hand to improve the efficiency of conversionor cost-effectiveness of photovoitaic cells but none is believedto be at ali closeto commercialisationasyet. Thesemethodsinclude: 0 Separationof incident light into wavelengthbandswhich are then applied to photovoltaic deviceschosenfor their sensitivity to eachband, giving overall efficienciesof up to 50%. ii) Vertical multi-junction cell, in which conventionalsilicon cells arestacked in such a way that light entervparallel to the junctions. Theseoffer the possibility of efficienciesin excessof 20% and it hasbeenclaimed that 147 such cells are not so sensitiveto reduction of performanceunder high temperatures,hencethey are better suited for usein concentratingphotovoltaic arrays of much higher potential efficiency than has so far been possible. iii) Thermophotovoltaic cell, in which sunlight is focussedonto a thermal massin an enclosuredesignedto maintain the thermal massat a temperature of 18OOoC.A layer of cooled photovoltaic silicon cells is arranged behind the thermal massand receivesradiation at the wavelengthcorrespondingto an emitter at 18OOW,which happensto be the wavelength to which silicon solar cellsare particularly sensitive.In this way the relatively broad solar spectrumis convertedwith reasonablyhigh efficiency to a much narrower wavelengthband at which it can be convertedwith high efficiency to electricity by silicon cells.An overall systemefficiency (solar radiation to electricity) of the order of 30 to 50% may thereby be achieved. If the expectedfuture cost reductions in conventionalphotovoitaic cellsmaterialisethen it is questionablewhether any of theseinnovations will be economical!y justified. 6.7 Memory Metal and other Solid-StateHeat Engines Various alloys, including one of nickel and titanium developedby the US Naval OrdnanceLaboratory called Nitinol, havethe property that they deform easily at low temperaturesbut return to their original shapewith considerableforce if heated.Varicus simple heat engineshave been demonstratedin which Nitinol elementsare lsed by being alternately heatedand cooled betweenwarm water and air. So far ths techno!ogy is at an early stageand low efficiencies havebeen reported. However,the rather small temperaturedifferencesrequired meanthat it may be possiblefor such a deviceto function with simple and low cost (per unit area)collectors(e.g. possiblya solar pond) which may offset the disadvantageof low efficiency. Differential expansionof dissimilar metal strips is also a potential sourceof power from a low temperatureheat source,but the efficiency is againlikely to be very low. Rubber and other polymers haverecently beenshown to havesimilar properties to Nitbinol and enginesusing these have been demonstratedin the laboratory. Howevertheir efficienciesseemlow and the are still at too early a stagein their developmentfor it to be clearwhether they may eventually be either technically or economically viable. 6.8 Osmotic PressureEngines In theory, thermal energycan be convertedto mechanicalenergyfor pumping by an osmotic process.In this process,a dilute solution, such asbrackishwater, is distilled usingsolar heat into separatesolventand concentrate(e.g. distilled water 148 and saturatedbrine solution). The concentratedsolution and solvent are then passed into two chambersseparatedby a semi-permeablemembrane.The resulting osmotic pressureset up acrossthe membraneis very substantial(up to 380 atmospheresin the caseof concentrated brine and distilled water), and this pressure could be usedto drive a turbine or other mechanicaldevice(seeReference37). Although the techniquesfor solar distillation are simple and well-established,it is not thought thar any serious investigationinto the feasibility of energyconversion basedon osmotic pressureshastaken place. 149 7. REFERENCES I. UNDP/World Bank/Sir Wiiam Halcrow & Partners/lntennediate Technology Dev.Grp. Ltd., “Testing % Demonstration of Small-ScaleSolar-Poweredpumping Systems:State-of-Art Report”, London, December1979. 2 SaMaar,G., & Mess,C.J., “An overviewof the water pumping equipment available to sma8 fanners” International Rice ResearchInstitute, Los Banos,Phihppine& 1979. 3. World Bank “Energy in the DevelopingCountries”, Washington,August 1980, fP 53). 4. Stem, Peter “Small-ScaleIrrigation”, I T Publications, London, 1979, (p 16 et seq) 5. World Bank “World Bank Development Report, 1980”, World Bank/Oxford University Press,New York, 1980. (p42-3). 6. Leach,Ceraid “Energy & Food Production”, International Institute for Environ& Development,London, 1975 (~9-15). ment 7. World Water,Ott 1980, “The Right to Rice”. 8. Smith, DouglasV., “Photovoltaic power in lessdevelopedcountries”, MIT., Massachusetts, 1977. 9. Langerhorst,J., et al, “Solar Energy,report of a study on the difficulties involved in applying solar energyin developingcountries” NetherlandsMinistry for Development Cooperation, 1977. 10. Tabon, Richard D., “The Economicsof WaterLifting for Small-ScaleIrrigation in the Third World: Traditional and Photovoltaic Technologies“M.I.T. Energy Laboratory Report M.I.T. - EL - 79 - 011, Cambridge,Mass.,1979. II. Makhijani, Aijun and Poole; Alan, “Energy & Agriculture in the Third World,” Ballinger,Cambridge,Mass.,USA, 1975. 12. Electrical ResearchAssociationLtd., “Electric PowerPlant International”, 1978 79 Edition, Leatherhead,Surrey, 1980. 13. Kooi, 14. French, David,; “The Economicsof RenewableEnergy Systemsfor Developing Countries”, a1Dir’iyyah Institute & USAID funded private report, WashingtonDC, USA. 15. Jansen,W.A.M., “PerformanceTestsof KeroseneEnginePumpsets”,Wind Energy Unit, WaterResourcesBoard, Colombo, Sri Lanka, 1979. ClarenceF., “Cost Comparisonof Dieseland Photovoltaic Pumpsfor ‘Action Ble’ “, REDSO/WA,(private communication, Dee 1980). 150 16. J.D.Walton,A.H.Roy and S.H.Bomar,“A State-of-the-Art surveyof solar powered irrigation pumps, solar cookers and wood burning stovesfor use in sub-saharan Africa”, GeorgiaInstitute of Technology,Atlanta, USA, 1978. 17. J.Giri and B. Meunier, “Evaluation des energienouvellespur le developpement desetats africains”, Ministry of Co-operation,France, 1977. 18. MNBahadori, “Solar WaterPumping”, Solar Energy 2 1 (4) p. 307, 1978. 19. J.T. PytIinski, “Solar energy installations for pumping irrigation water”, Solar Energy 21 (4) p 255, 1978. 20. E.H. Lysen, “Solar pumps”, TOOL Compendium 5, April 1979. 21. P.L. Fmenkel (editor), “The PowerGuide”, Intermediate Technology Publications, 1979. 22. D.L.Larson and C.D. SandsII, “Considerationsin using solar energy to drive irrigation pumps”, Proc. of the International Conferenceon Energy UseManage ment. PergamonPress,1977. 23. J.P. Durand, M.G. Clemot, J.P. Girardier and M.Y Vergnet, “Utilisation of solar energyin the developmentof arid zones:solar water pumps”, Technology for Solar Energy Utihsation UNIDO Developmentand Transfer of Technology SeriesNo. 5, 1978. 24. W.K. Kennedy, “The appropriatenessand implementation of solar energyin rural developmentprogrammesin developingcountries”, Proc. of the UNESCO/WMO Symposium,Geneva,WMO477, September1976. 25. US Department of Energy Target Costsfor Photovoltaics,quoted by R.G.Fomey, “Photovoltaics in the USA - a progressreport”, Proc. of One-dayConferenceon Photovoltaic Solar Energy Conversion,UK-ISES, September1979. 26. Maycock, Paul- Presentationat 15th IEEE PV SpecialistsConf., Orlando, Florida, May, 1981. 27. Gay, CharlesF., “Solar CeUTechnology: an assessment of the State of the Art”, Solar EngineeringMagazine,Dallas,Texas,Vol. 5, No.3, March 1980. 28. R.W. Mathn, “Design optimization and performancecharacteristicsof a photovoltaic micro-irrigation system for use in developingcountires”, MIT Lincoln Laboratory, Lexington, USA, July 1979. 29. D.Campana,A. Castiel,A. Perez,J.A. Roger,C. Dupuy, P. Labit and M. Lepert, “Realisation and testing of a pumping systempoweredby solarcells”, Proc. of the UNESCO/WMOSymposium,Geneva,WMO477, September1976. 30. Butti, K. and Perhn,S., “A Golden Thread: 2500 yearsof Solar architectureand Technology”, CheshireBooks/Van Nostrand Reinhold, 1980. / 151 I 31. Sandia National Laboratories, “Concentrator Collector Results, Report SAND 8Cl-0865,Albuqurque. NM. USA, 1980. 32. R.L.Wood and E.A.Platt, “Shallow sohn pond powered irrigation pumping: a feasibility study”, Proc American Section ISES Annual Meeting,P 4 I - 45, Denver, USA, August 1978. 33. W.T.Beale,“A free cylinder Stirling enginesolar poweredwater pump”, Sunpower Inc., Athens, Ohio, USA, 1979. 34. D.R.Gedeon, ‘The optimization of Stirling cycle machines”, Sunpower Inc., Athens, Ohio. USA. 1979. 35. G.G.Puri, “A reflector concentrator modified Stirling engineunit for farm power needs”, Pmt. International Solar Energy Society Congress,New Delhi, 1978. 36. V. Baum and N. Ovezsackatov,“Solar thermoelectric generators,attainable efftciency and perspectives”,Proc. International Solar Energy Society Congress, New Delhi, 1978. 37. “Application of Solar technology to today’s energyneeds”,Office of Technology Assessment,Congressof the United States,Washington,USA, 1978. 38. DoorenbosJ., and Pritt W.O., “Crop Water Requirements” FAO Irrigation and DrainagePaperNo. 24.1977. 152 153 APPENDICES 1. Prdimhry Estimates of Costs of Solar Pumping Systems in Developing Countries 2. General Recommendetions systrms 3. Executive Summary of the Project Report 4. Objectives of and Prepration for the Development of SmaCScale Solar Pumping for phase II of the Project 154 APPE.NDIX 1 Preliminary Witnates of Costs of Solar Pumping Systems in Developing Countries $y:;;,“, 155 APPENDIX 1 PRELIMINARY ESTIMATES OF COSTS OF SOLAR PUMPING SYSTEMS IN DEVELOPING COUNTRIES 1. Beckground With the continuous escalationin the price of fossil fuels, and the anticipated global shortageof oil, the attractions of solar energyhavemagnified enormously over the last ten years.The American governmentgavea target in 1979 that renewableenergysourcesmust satisfy 20% of total U.S. energyneedsby the year 2000, and naturally this gavemuch stimulus to the solar industry in the USA. However,all this activity is not necessarilybeneficial so far asthe needsof develop in countries are concerned.The US programmesare of national, political, and military sigmticanee,and while the economic aspectis obviously important it is not necessarilyall important. The majority of the end uses,and the vastpurchasingpowersof the U.S. market are quite different from those of developing countries. Therefore much of the developmentis high technology, large scale, sophisticated,and in a country with an almost unlimited manufacturingbaseplus the highly skilled maintenancefollow up required. Developingcountries offer a complete and stark contrast to the USA. Technology is limited, and so is the supply of technicians,the manufacturingbaseis small, only fairly standardraw materiaisareavailable(with imports of specialitems difticult due to lack of foreign currency), quality control at manufacturesnd after sales ma+enance follow up is likely to be inadequate. Manufacture of solar pumping systemsin the developingcountries, beneficial though it might be, is therefore by no meansan automatic spin-off from work carried out in the industriahsedcountries. Fortunately there is a plus to this situation aswell asa minus. Appropriate solar pumping system technology must aim for maximum simplicity, which in turn can lead to lowei prices and should give better reliability. 2. Cost end Market Given a low enoughcost there is undoubtedly a very largeworld wide demandfor reliable small-scalesolar-poweredpumps for irrigation and water supply. While first cost is not the only factor in the overall economicequation it is almost certainly the most important. As mentioned earlier this can work in favour of the Project sincelow fmt cost impiies simplicity and small-size,which in turn implies a design suitable for manufacturein the developingcountries. 3. - Estimate of costs of quantity production of a 200 watt PV system The simple type of sythen is assumed: Array/DC Motor/Centrifugal Pump A maximum power point tracker (MPPT) is also consideredin casetheseturn out to be ’ cost-effestive.Cost estimatesfor quantity production in a developingcountry in 1984 are made for eachcomponent as follows: Modulesfrom USA at 6 5 per peak watt (At a later stagecells may be imported and built into modulesin the developing tiuntries -though the valueaddedwill be modes;) 50%addedfor packing,air freight, import hc’:‘. and iocal fabrication into arraysincluding provision for 2 axis adjustment. This gives$7.50 per watt. .Assuming50% wire to water efficiency, array must produce 400 Watts. cost = $3000 MPPT Initial import of 10 complete units plus 100 setsof componentsto the develop ing country. Deletion program increaseslocal content to 100%as local componentsbecome available. Cost taken as$200 w DC Motor, Developingcountry will probably haveto import (initially at least) copperwire for windings, ball bearings,seals,permanent magnetsand armature stampings. Hopefully commutation will be brushless,position unclear at the moment. Shaft, casing,and variousother parts to be madein developingcountry. Armature winding and motor assemblylocally. Total cost taken as$250 157 centrifugal Pump Main casings(cast iron) and shaft producedin developingcountry. initial import, but later local production, of impeller (sintered metal or plastic), and shaft seal(spring loaded graphite/steel),and plain bearings(probably sintered metal). Total cost taken as$300 Thus the target cost for the complete systemis $3750 It should be stated that: (a) The significanceof the array cost as80% of the overall price build up is apparentevenwith cells purchasedat $5, Y pk (b) No attempt hasbeen made to estimateinstallation costssincetheseby their nature are site specific. 158 4. Estimate of costsof quantity production of the 200 Watt thermal system A conventionaldesignis assumedusinga simple line-focussinglow concentration tracking collector with the Solar Pump Corporation systemcomponentstaken asmodelsfor casting the balanceof the system.Cost estimatesfor quantity production in the developingcountry in 1984 for eachcomponent asfollows: (9 Collector Copper pipe, reflectors and tracking unit for concentrator probably imported. Steel framing, insulation and overall fabrication assumedavailablein developing country. Given someefficiency increasefrom developmentwork up to 1984,areafor 200 watts peak pumping power could be about I Om’ Taking $150 per squaremetre, (comparedwith current U.S. manufacturer’sprice of S 120/m’ including tracker) Cost = $1500 (ii). Main Frame Fabricated from local standardstock steel, and including provision for mounting of collector and tracker. Cost taken as8400 (ii) Freon Engine & Valve Main rubber diaphragm,various seals,and some small componentsimported, initially at least. Main housing, piston rod, piston rod guide etc, and most kralveparts cast and machinedin developingcountry. Cost taken as$350 (iv) Condenser Madein developingcountry Cost taken as$250 159 69 Freon Pump Sealsimported, o-~‘~~~iselocally made, Cost taken as$100 w Pressure Relief Valve Locally made Cost taken asSSO (W Various Levers etc Locally made Cost taken as$100 (viii) WeU Head Assembly Locally made. Cost will clearly dependon cylinder sizeand depth of well. For a IO metre depth it would seemnot unreasonableto take cost as :$400 (W Freon Taken as$100 co Piping & Assembly Taken as$250 Thus the target cost for the complete systemis S3600. It should be noted that:(a) SPC have “guestimated” a selling price of S 4000 for quantity production in Las Vegas of their flat plate collector, but otherwise similar system. 160 (b) No figure is included for site preparation, foundation work, etc. The assemblyfigure (8 250) coversthe assemblyand pipe connection work. Cd It is felt that considerabledesign,development,and test work will be necessarybefore any sinall thermal system is suitable for quantity production, either in developing countries or in industrialisedcountries. 161 APPENDIX 2 GeneralRecommendationsfor the Developmentof Small-ScaleSolar PumpingSystems APPEMl% 2 GENERAL RECOhlMENDATlONS FOR THE DEVELOPMENT OF SMALLSCALE SOLAR PUMPXNGSYSTEMS* 1 Applications Small-scalesolar pumps for irrigation purposes are more likely to be successfulif they are: 0 applied in aresswhere irrigation has been traditionally practised, so that it is only the use of the novel pump technology and not irrigation per se that has to be taught 0 sized for small land holdings of the order of 0.5 ha.. for which they are most competitive with the alternatives 0 first promoted in areaswhich need to be irrigated over most of the year and where multiple cropping might be prsctised io order to achieve a good year-round load factor for the solar pump 0 preferably restricted, at least initially, to pumping static heads of no more than around 5m (and IOm at the very highest), implying a system output of the order of 100 to 300W peak of hydraulic power for a land area of 0.5 ha. 0 usedgenerally in areaswhere the provision of fuel for enginesis proving increasingly difficult and costly and where mains electric power is absent. It should be noted that the first costs of the more cost-effective of the present generation of pumps are still at least double that of systemswhich would represent an economic investment when used for irrigation. The aspectis discussedfully in the Technical and Economic review. Small-scalesolar pumps should also be considered for application to non-irrigation purposessuch as village or livestock water supp::zs, which offer a better load factor (demand being more or lessconstant all the year around). Another point of fundamental importance is that the price which water can command for domestic purposesis considerably higher than for irrigation. It was beyond the scopeof the project so far to investigate these other end-usesbut it is expected that solar pumps would be viable at a much higher head (perhaps up to 20m) for this kind of application. 2 Photovoltaic System Spe&ication As a result of the studies completed under this project, it is possible to produce an outline specification for a small-scalePV pumping system for irrigation at low heads, that ought to be competitive with small engine powered pumps in many parts of the world. It must be stressedthat this is not the ultimate type of design that is to be recommended, but the logical next step incorporating the principal lessons learnt so far in using present day equipment. It is to be expected, of course, that innovative developments in the technology will introduce new options in the near ftiture that are no? considered here, but which have been discussed in the Project Record and in the Technical and Economic Review accompanying this Report. l A copy of Chaprer I 1 ‘Design Recommendations’ from the ‘Project Report 163 iii) System Efficiency Requirement Optimum system efficiency targets should be set as follows for the next generation of pumping systems: array cells connections motor pump pipework 11% 95% 85% 55% 95% total for system 4.6% at 25’C (10% at NOCT) (based on array cell area) This is equivalent to the very best systems so far tested. Any power con-. ditioner or other such accessories that consume power, should provide benefits to more than compensate for their cost and power consumption. iv) _ Design Point to be observed In preparing designs to meet performance specifications of the type outlined in 11.2.2 and the target efficiences set out in 11.2.3, it is recommended that manufacturers pay due regard to the following practical design points. These are not listed with the intention of inhibiting future development in any way, but as a convenient resume of the main lessons learnt during Phase I of the Project. a) Genera! Requirements 0 0 0 0 0 the system should either be fully portable and mounted on skids or wheels or it should be fixed. In the latter case the array should ba mounted at least 1.5m above ground level and bolted to not more than two concrete foundation blocks; the holding down bolts to be positioned to an accuracy no better than 25mm, to allow for errors in the positioning of the foundations. systems should be designed to survive particular local climatic conditions. In areas prone to typhoons or hurricanes it may be necessary to provide for the rapid dismantling and removal to safety of the array. pumps and motors should be provided with sun-shades whenever possible, but be well ventilated if air-cooled. materials exposed to solar radiation (such as plastics) should have proven durability. modules should be individually packed to avoid damage in transit. No single package for the system should exceed Im x 2m or a weight of 50kg. Safety of contents should not be dependent on being stored any particular way up or on receiving special handling treatment. 164 0 0 detailed instructions should be provided, in the local language, for the correct assembly of the system. Also included should be a components list, maintenance and operating instructions and guidance on how to achieve the best output (eg, not using small bore pipes or restricting fittings, not allowing shadows to fall on the array). All documentation should be written in simple terms. the need for routine maintenance lubrication and readjustments (eg, belt-drives and waterseal packing) should b,e minimised and avoided if possible. Array and Module Requirements 0 0 lifetime guarantee of say 5 years against faulty quality with no more than 10% degradation of performance to be accepted. small modules (not more than about 20W nominal rating preferred). 0 laminated glass or an impervious and ultra-violet resistant plastic cover. 0 optimum module cell efficiency exceeding 11%. 0 redundant interconnects between cells. 0 no ah gap between cell encapsulant and cover glass. 0 generously sized (brass) terminals with grip screws, (plated steel terminals not acceptable) or appropriate plug and socket connectors. 0 0 weather-sealed terminal boxes behind array. array frame should be capable of being manually tracked on an equatorial axis at intervals through the day. 0 array inclination should be adjustable and preferably engraved with correct angles for different season for the region. 0 0 cables should be generous in cross-section, to limit resistive losses at full power to no more than 5%; blocking diodes (if required) should result in no more than 2% loss at full power. array nominal voltage under peak sunlight conditions should be around 8OV: higher voltages will not be safe, while lower voltages will mean higher currents and larger resistive losses. 165 0 electronics power conditioners (or other electronic ‘circuitry) should be fully ‘tropicalised’ and should use components to tropical ambient temperature specifications. Full protection by automatic cut-out against excessive temperature, voltage or current, (prefer automatic reset). 0 quality assuranceand testing to be to a satisfactory standard and specified by the manufacturer. 0 cl module or array performance to be specified on the basis of tests to a stated international standard. The module/array area to which array efficiency is referred is to be clearly defined. Subsystem(motor-pump unit plus fittings) Requirements dc permanent magnet motor preferred unless altemative shown to be of comparable efficiency. fa&safe brush gear (motor stops when brushes too worn), or electronically commutated motor, (see’ electronic requirements shove). If brushes are used, life of 4000 hours minimum required between changes. Sufficient spare brushes for the expected life of the system should be provided. 0 generously sized brassterminals with grip screws(plated steel terminals not acceptable) or appropriate plug and socket connectors. full load motor efficiency should exceed 85%. half load motor efficiency should exceed 15%. motor bearings and other components should be sized for a life in excessof 10,000 hours. submersible units are preferred to eliminate suction problems. if motor not submersible, coupling between motor and pump should permit significant motor and pump angular and/or parallel misalignment. Motor and pump bearings should he entirely independent, except in the caseof integral submersible motor-pump units. 0 thermal cut-out required on motor, unless motor can sustain continuous stalled conditions with the maximum array current. 0 pump optimum efficiency should be in excessof 50%. 0 pump should be capable of self-priming in the event of a leaking footvalve (where fitted). 166 impeller material, clearances and passagesshould be suitable for use with watercontaining suspended silt and/or corrosive salts. Open impellers are preferred. where necessary, a suitable strainer for larger particles should be provided and to be sized so as to have negligible effect on performance while clear. pumps should not normally weigh in excess of about 30 kg for low head irrigation application, and should preferably be no heavier than 20 kg (assuming peak flow rates in the range of 2-5 l/s). pump should have bearings sized for 10,000 hours of operation and be supplied with spare seals or any other consumables to cover that period of operation; prefer ball bearings with grease seals. pump characteristic should permit stable operation at sub-optimum speeds. pump should be capable of running dry without serious damage lf not submersible; alternatively, fail-safe method of protection from running when dry should be provided, and seals designed.to suit. pipework and fittings should be correctly optimised for minimum system cost, rather than minimum pipework cost; all pipework should be supplied with system to length specified for head and site. 3 ‘Ihenna System Development FolIowing the laboratory tests of thermal systems, the system design studies, and the installation in the field of a complete system, definite recommendations for the development of small-scale solar-thermal water pumping systems can be made. The continued development of solar-thermal systems should be encouraged in particular to demonstrate small-scale systems utilizing concentrating solar collectors with organic Rankine cycle engines or high temperature air Stirling cycle enpines. Existing working systems have been shown to have the potential for quantity production at costs comparable with the present costs of photovoltaic systems. Further development of these type of systems is justified if significant future reductions in the cost of photovolaics is in doubt in which case development should be encouraged to improve the reliability and performance in order to achieve designs more appropriate to the proposed use. Further work is required in order to broaden the limited scope of the present thermal design studies and in particular to investigate the designs that have been shown to have promise for low cost manufacture. 167 Specific Capital Cost analysison small thermal systemsgave$1.5/kJ per’ds.yfor a Freon vapour Rankine cycle engine with a one-axis tracking parabolic trough (line-focus) colIector. A high temperatureair Stirling cycle enginewith a pointfocus, tracking parabolicdish collector or central receivercollector also should be further investigatedas the Specific Capital Cost was determined as being even lower. Componentsin this type of systemare howeverlessdevelopedthan thosein line-focuscollector Rankine cycle enginesystems Componentsthat may form part of an improved soiar-thermalsystemshould be evaluatedand tested. Fur example an inexpensivereliable tracking mechanism for a s&r collector needsto be investigated. Improved systemswIIl need thorough laboratory tests followed by field trials. This wIII alsoimprove the data basefor developmentof the mathematicalmodel. The sizing of thermal systemsneedsfurther study to improve Specific Capital Costs.This is more size dependant for thermal systemsthan for photovoltaics. Doily output m3 . - I I b+ Hwd m Daily output Cm7 at head bm. b. i r b Daily irradiation Fimme 5% 7 characteristics kWh/m2 for uhotovolfaic 169 APPENDM 3 Rxeeutive Summery of the Project Report I’ 170 Smali-ScaleSolar-PoweredIrrigation PumpingSystems: PhaseI Project Report EXECUTXVESUMMARY 1. INTRODUCl-ION 1.1 Purposeand Background The Project was funded by the UNDP and executed by the World Bank. The Consultantsappointed to implement the technical work were Sir William Halcrow &Partners working in associationwith the intermediate Technology Development Group (both of London, UK). PhaseI of the Project commencedin July 1979 and lasteduntil May I98 1. The main purposeof PhaseI of the Project hasbeento demonstrateand evaluate the useof solar energyfor powering small-scalepumping systemsto be usedfor irrigation on typical small land holdings in developingcountries,with a view to recommendinghow the technology should develop. In this connection, “small” land holdings refers to the millions of intensely cultivated land holdings farmed by poorer farmersin many developingcountries, the majority of which haveareasof around one hectareor lessrequiring, typically, about 50m of water per day per hectarein the irrigation season.The hydraulic power output required to pump suchdaily volumesfrom depthsof typically Sm, will be of the order of IOO-300W. In’order to evaluatepumps for this duty, considerablepractical work wasrequired, including field trials of systemsand laboratory testing of subsystemsand components,followed by an analytical systemdesignstudy. As a result of this work, the likely performanceto be expectedfrom small-scale solar pumpsasthe technology matureshasbeendeterminedwith someprecision. The advantagesand disadvantagesof different technical options havebeenstudied and indications obtained about the desirablefeaturesto be developedfor smallscaleirrigation pumping systems. Some forecastsof the likely price trends and of the factors that affect the economicsof suchsystemshavealso beenmade. During the courseof the Project, much useful experiencewasgainedabout the proceduresnecessaryfor testing the equipment in the field and laboratory and for data collection and reporting. Valuable lessonswere also learnt about the collaborative relationshipswhich needto be establishedwith the co-operatingagencies in the host countries and thesewill be useful to future phasesof this Project. 1.2 Reports This Volume containsa description of the project and the substantiveconclusions which were reachedasa result of the work done in PhaseI. An accompanying Volume reviewingthe technical state-of-the-art,possiblefuture developmentsand 171 the generaleconomicsof small-scalesolar pumps, is entitled “Small-ScaleSolarPoweredIrrigation Pumping Systems:Technical and Economic Review”. PhaseI of the project conveniently breaksdown into three principal componentsfield trials, laboratory testsand systemdesignstudies-andthis is the generalformat of the Project Report and of the summarywhich follows. 2. FIELDTRIALS 2.1 Objectives The main objectivesof the field trials were: I. To obtain fmt hand, reliable and objective performancedata for existing systemsunder field conditions as a prerequisitefor assessinghow the technology should be improved. 2. To gain experiencewith the managementof field testing, data collection and analysisand reporting. 3. To demonstrate the technology in countries typical of those in which solar pumping may havewidespreadfuture application asa preliminary to technology transfer, and to highlight the aspectswhich most urgently require attention. 4. . To providedata for the validation of the mathematicalmodel built aspart of the systemdesignstudies. The DNDP selectedMali, Philippinesand Sudanto participate in the field trials. The Consultantscollaborated with the Solar Energy Laboratory in Mali, the Center for NonConventional Energy Developmentin Philippines,and the Institute of Energy Researchin Sudan. 2.2 Selection of Equipment The Consultantsreviewedthe availability of small-scalesolar pumps suitable for the purposesof the Project. 250 questionaireswere sent to potential suppliersthroughout the world. However, although many countries have significant R & D pmgrammes.few suppliers were in a position to offer adequatelydeveloped systemsto satisfy the selectioncriteria and within the delivery scheduledemanded by the Project. The selectioncriteria required equipment with the correct delivery for the head specified, typically in the range I to 3 litrelsec through a static headof 5 to IOm under peak sunlight conditions. Also, systemswere required to be robust and 172 practical. have low maintenance requirements, be efficient and preferably have some promise of capability of manufacture in developing countries. The required delivery time was I3 weeks from time of order. Only 13 suppliers were able to offer equipment that appeared suitable and a short list was prepared for approval by the World Bank. Finally, IO different systems were purchased, nine photovoltaic (PV) and one thermal. One PV system was duplicated and so 1 I systems in all were purchased for the field trials. In addition, permission was given to monitor an existing system already installed and working in Mali, but not purchased by the Project. The 12 systems monitored, their respective costs (including air freight to host country), locations and the volume of data collected for them are indicated in Table I. 2.3 Field Programme and Progress The field programmes in each country were supported by Resident Engineers (RE’s), one in each country, posted by the Consultants to assist the national institution install the equipment and to advise the local engineers and technicians in its use and in the monitoring procedures. The RE’s had previously visited the manufacturers of the systems they were to install to inspect the equipment prior to despatch and to receive instruction and advice. The principal parameters measured in the field were concerned with performance assessment,both instantaneously through continuous monitoring of key variables (such as irradiance in the p!ana of the array, photovoltaic array electrical power output and pumped warer rlow rare) using cnart recorders and cumulatively, by daily readings of inputs and outputs (such as solar energy input during the whole day, electrical energy generated and total quantity of water pumped) registered on totalizers. In the event, the recording of field data presented some difficulties: technically with the pumping systems and to a lesser extent, with the instruments; and logistically with late deliveries by the manufacturers, damaged key items, and difficulty in arranging transport to the sites. The short time scale of Phase I also created other problems: equipment could only be ordered in January 1980, testing commenced mainly between June and September 1980, and the RE’s left their countries by September 1980 (returning briefly later in December 1980 and/or January 1981). Despite these matters, the field trials programme was substantially completed by the end of Phase 1. Of the 12 systems monitored, 10 yielded sufficient continuous data to determine their performance, while two systems could not be tested at all; in one~case(a PV system) this was because of delays in arrival of parts and damage in’transit! while in the other (the thermal system) technical problems prevented the pump from running for periods long enough for its performance to be determined (though these defects were of a kind which, given time, should be solvable) Of the IO PV systems from which data were obtained, one borehole pumping system performed almost faultlessly for the whole of the trial period. A number of problems afflicted 173 other PV systems including, for example, two systems with poor suction performance (even under high level; oi irradiance), two systems with unreliable footvalves (leading to pumps being difficult to prime or running dry), two systems with wrong wiring or poor connnections (leading to low power output, from the arrays), one system with an impeller binding on the pump casing and two systems with unreliable electronics (leading to complete failure). Some faults were repaired in situ, while others required replacement part:. The proolrms involving poor suction and impeller binding were not resolved within the time scale of Phase I of the Project. 2.4 Conclusions from Field Trials Table II indicates the principal results obtained from testing all the systems. The maximum instantaneous system efficiency (based on array area) recorded was just over 3%. However, most of the adequately reliable systems were typically 2% efficient and the poorer systems only returned optimum efficiencies of around 1%. Referenced to gross cell area, the maximum system efficiency was just over 4%. Clearly, there is considerable variability in efficiency between different systems, and since the efficiency dictates the size of array necessary for a given output, and array costs dominate, overall efficiency has a major effect on equivalent annual costs. From an operational viewpoint, the main conclusion was that pumps have to be self-priming, that is, they must be able to start pumping without any need for operator intervention. Non-self priming pumps, if emptied of water due to, say, a leaking footvalve, cannot refill themselves and therefore will simply run dry, overheat and may suffer serious damage. Wherever possible, centrifugal pumps should be of the immersed type and so dispense with troublesome and energy consuming footvalves. If surface-mounted self-priming centrifugal suction pumps have to be used, care should also be taken to see that the suction head remains moderate. Numerous relatively minor problems Yere experienced with many of the systems; for example: incorrect wiring supplied terminals that did not readily give good electrical connection faiiure of electronic circuitry (due either to overheating or to overload) possibility of safety hazard due to dangerous dc voltages broken module cover glasses(both in transit and on site) suction pi$ework trapping air in cavities footvalves jamming or leaking inadequate packing for shipping It was felt, however, that these problems were not fundamental to the technology and can be overcome during the normal course of development as the technology matures. Certainly, the systems have the potential for reliable operation with minimum maintenance. Regular maintenance jobs should be minimized and made easy to carry out. 174 3. LABORATORY TESTS 3.1 Objectives The main objectives of the laboratory testing programme were: (1) To determine independently the true performance characteristics of selected solar pumping subsystems and components under controlled conditions at full and part-load. (2) To provide data for the system design studies to enable improved solar pumping systems to be developed. (3) To help identify the causes of good or bad field-tested system performance. (4) To provide limited indications of the basic reliability and durability of certain components. Generally, the component testing programme was designed to investigate performance (i.e., output and efficiency characteristics) rather than reliability or quality. Hence, single items were tested in the case of PV sub systems (motors and pumps) and of thermal systems, but for the reasons given below, five examples of each selected PV module were tested. The testing programme breaks down into three principal areas of investigation: PV modules, PV subsystems (motor and pump) and thermal systems. 3.2 PV Modules The choice of PV modules was limited to those which had not been independently tested and publicly reported and which displayed interesting features - technical (e.g. high packing density for cells), commercial (e.g. low cost) or others, such as a product made in India which was indicative of what might readily be manufactured in a developing country. All cells were of the mono-crystalline silicon type. Since these are particularly expensive components which are required to be durable, robust and long-lasting, five examples of each module were purchased for a more detailed investigation of quality and performance. Details of products tested are given in Table III. The modules were performance tested independently by the Royal Aircraft Establishement (RAE) in the U.K. and by the Jet Propulsion Laboratory (JPL) in the U.S.A. RAE carried out oerformance tests and ultra-violet accelerated ageing tests and JPL carried out their standard performance and durability testing (a routine generally applied by JPL for the US Department of Energy to most American PV modules). Both testing organizations carried out visual inspections and reported on the condition of the modules before and after testing. The results obtained by each of the organizations for the maximum power output of the same module were very close (average value differed by 0.5%). 175 3.3 PV Subsystems The University of Reading in the UK tested pump and motor units from all the PV systems purchased for testing in the field. In addition a selection of four further pumps and two motors were tested. The extra pumps were chosen as representing generic types of pump not otherwise included in the programme but of possible interest for the future design of pumping systems. These included a free-diaphragm pump, a rotary screw (or progressive cavity) pump, a piston pump and a selfpriming centrifugal pump. The extra motors included a high efficiency industrial permanent magnet dc motor, similar to those used on most of the subsystems, and a reciprocating linear actuator. Details are given in Table III. The main requirement of this testing programme was to obtain performance data vital for the system design study over a wide range of full and part-load conditions and typical of those existing in solar pumping applications. Components were also visually inspected after testing to determine their general quality and suitability for inclusion in small-scale irrigation pumping systems, 3.4 Thermal Systems Virtually all thermal systems were of prototype status - only one such system could be purchased for field trials and even this was not a mature product. The only system which could readily be transported for laboratory testing at the University of Reading was a Stirling cycle solar pump, under development in the USA. Hence the only generally practicable way to gain an insight into thermal system performance and potential was to witness testing at the manufacturers’ own test facilities; hence arrangements were made for the Consultants’ engineers to monitor tests of three systems under development in Israel, Germany and the U.S.A. Although other systems were known to be under development, it was only possible to make arrangements for those listed in Table III. 3.5 Conclusions of Laboratory Testing The objectives of the laboratory testing programme were generally achieved, with reports of completed tests on all of the equipment. Modules The performances of the PV modules were consistently below their manufacturers’ specifications, by from 2%0/oto I6H%. This has serious implications in view of their high cost (S IO to S20 per peak Watt) and because accurate performance knowledge is necessary for good system design. Cell efficiency varied by over 20% between the best and worst products tested. Although only one PV module actually failed under the durability testing programme, all products displayed minor flaws or design faults and one or two had potentially serious shortcomings, Although it is considered that the levels of efficiency ( IO%+). high quality and long life often claimed for PV arrays in the technical press and in manufacturers’ literature are obtainable, such a standard of performance of most currently available products cannot be taken for granted at present. 176 Motors The dc motors tested were all permanent magnet machines* of generally high efficiency. Nevertheless a spread of about 15% in optimum efficiency (75 to 87%) was found between the best and the worst performers. This difference can be worth more than the total cost of a motor in terms of extra array costs (at present day prices). Clearly, PV solar pumping systems should use motors of better than 85% optimum efficiency, so long as the cost of PV arrays is dominant. Considerable variability in pump performance was revealed. The best centrifugal pumps were over 50% efficient under optimum operating conditions, while many were only 30 to 40% efficient. A few were less than 20% efficient. It is clear that the choice of pump can be perhaps the single most influential factor in good smallscale solar PV pumping system design. From the system design studies it was found that overall system efticiencies for systems using centrifugal pumps were sensitive to head variation, and it is clearly important to seek pumps whose drop in efficiency when operating away from their optimum head is as small as possible. This objective will help to maintain acceptable performance for situations where the actual head does not coincide with the optimum for the pump or where the head varies either seasonally or due to well draw-down. Nevertheless, centrifugal pumps appear to be the most promising type of pump for the low head applications (< 1Om head) demanded for irrigation, although selfpriming capability is essential for practical field use. Positive displacement piston pumps offer good performance (better than 50%) at higher heads Q 10m). They are also relatively insensitive to variations in head or to operating off their optimum head. Thermal Systems The performance of the Stirling engine pumping system was disappointing and it failed mechanically partway through the test series. The other systems, all Rankine cycle using Freon type working fluids, performed in line with our expectations and gave overall efficiencies of between 1% and 2%. Examples seem some distance from commercial viability (with the possible exception of the system purchased for field testing), but our system studies showed that this technology could be competitive with PV systems, given suitable development and high volume production. The system design studies showed that thermal systems would be likely to improve in both efficiency and in cost-effectiveness if tracking concentrating collectors were used instead of the fixed flat plate collectors of the Rankine cycle systems so far tested. Consideration would need to be given to the maintenance requirements which this would impose. * One was of the brushless electronically commutated type 4. SYSTEM DESIGN STUDIES 4.1 Objectives The purpose of the system design studies was to investigate whether, and in what ways, small-scale solar pumping systems might be improved for irrigation purposes and to see whether the specification of an improved system could be developed. This was achieved by examining the cost-effectiveness of a number of technically Feasiblesystem options under a variety of operating conditions. For all systems, computer-based mathematical modelling techniques were used to simulate the performance and cost of the main components of both PV and thermal systems. 4.2 Models The PV systems model was developed using data from the laboratory testing programme, validated by comparison with the field performance data obtained For complete systems. Factors that were investigated are listed in Table IV. A special parameter, the Specific Capital Cost (the capital cost per unit energy output per day in US S per k.I) was used to assessthe cost-effectiveness of the alternative system options examined. For this purpose “normalized” costs were used to give a measure of the capital cost of a different systems, excluding arbitrary factors which may influence the price of individual items. A simple thermal system model was also produced to investigate the general merits of different thermal system concepts (i.e., the use of different types of solar collector, thermal cycle, working fluid and expander). The Specific Capital Cost was also used to express and compare the cost-effectiveness of thermal systems. Finally, a simple economic model For PV systems was constructed to investigate the sensitivity of typical PV solar systems to variations in selected economic and technical parameters (discount rate, life, differential movement in real prices) and to compare solar pumping system costs with those of small engine-powered pumping systems. The model calculated the present value of the various cash flow streams. This work is reported in detail in the separate “Technical and Economic Review” volume. 4.3 Conclusions PV systems Significant performance and cost-reducing improvements appear to be readily obtainable with PV systems at the existing level of technology developed from~ those tested under the programme so far. In particular overall efficiency can be improved by: 0 the use of pumps whose efficiency change with head over the likely working head range is as small as possible. 0 optimisation of the proportion oFPV cells that were connected in parallel and series within the module. (One system supplied was almost perfectly optimised, but with another an improvement of 13% in overall system efficiency could be achieved). 178 0 varying the array power to find the optimum value: in one case the Specific Capital Cost dropped by 19%when the array power was increased by 60%, due to a substantial increase in output and efficiency. 0 the use of a Maximum Power Point Tracker (MPPT)*. This was investigated and may have little or no benetit on well-matched systems but would be expected to improve less well-matched systems under varying climatic and head conditions. Pumped daily water output was increasedtypically by 15% on a clear day and by 25% on a hazy day through the use of a MPPT. 0 using movable or tracking PV arrays. The increasein output obtained by a perfectly tracked array (over a fixed array) was compared with the increasesobtained by arrays reoriented to preset positions once, twice or three times in a day and reset in the evening. It was found that reorentation twice daily allows the use of 95% of the energy received by a continuously tracked array. It is doubtful therefore whether the extra cost and complication of a continuously tracking system can be justified for applications such as inigation. 0 care in specifying pipework for systems. The pipework necessaryto connect the pump with the source and to deliver water to the point of application is a simple, but vitally important, part of the solar pumping system. The Specific Capital Cost was minimized for the particular conditions of the tests: it is unlikely that a pipe diameter less than SOmm should ever be used. The use of inadequate pipe diameters can dramatically reduce the system efficiency and increase the Specific Capital cost. The combined result of a number of these improvements for one system is shown in Table V. In general, the Consultants consider it is feasible to produce small-scalesolar pump ing systemshaving overall instantaneous system efficiencies above 4% (some of the systemstested were only around 2% efficient). Specific Capital Costs of around $2.5 kJ** per day should be possible (at present day prices with +rrent zchnology) compared to a range of 63.1 to S9.8 obtained from model tests on the systemspurchased. If PV array prices fall to about 50% of their current lowest level (i.e. reach S 4/W peak) then the Specific Capital Cost of g 2S/kJ per day will reduce to about g 1.0 yielding water at about one third the cost of the most costefficient systemscurrently available. Thermal Systems Data on thermal systems were restricted in scope and quality compared with the PV data, but neverthelessclear conclusions emerged from the studies which analysed over 30 different system options. The principal one is that high temperature * A Maximum Power Point Tracker (MPPT) is an electronic control device which continuously adjusts the array voltages to an optimum value to maxim&e the power output from the array. Its main benefit is increasedsystem efficiency and output, but it also consumesa proportion of the power produced by the array and it can be an expensive component adding significantly to system first costs. ** The unit kilojoule has been used because it is a basic SI unit of energy and becauseit gives convenient dollar numbers. There are 3600 kJ in one kWh. 179 thermal systems using concentrating collectors appear to be significantly more cost-effective than low temperature flat plate collector systems, mainly on account of the much smaller collector areas required to yield a given output. The least cost-effective approach appears to be precisely the one favoured by most current manufacturers (fixed flat plate collectors); the indications are that a system using a linear-focussing equatorially-tracking collector would be about 30 to 40% less expensive, lvhile even higher concentration through the use of a power tower or point focus two-axis tracking collector could result in some slight further improvement in cost effectiveness. Some uncertainties lie in the assumptions concerning the likely costs of tracking mechanisms, but the results are nevertheless significant enough to justify serious study of the use of concentrating and tracking thermal concentrators, and the more complex maintenance requirements this would impose. Specific Capital Cost analysis on small thermal systems gave S2.8/kJ per day for a typical flat plate collector system (similar to the one tested) and S I.5 for a Freon vapour Rankine cycle engine with a one-axis tracking parabolic trough (line focus) collector: these costs appear to be competitive with PV systems. Thus it would be premature to dismiss thermal systems at this stage. 5. RECOMMENDATIONS 5.1 General 0 0 5.2 Solar pumping for irrigation is most suitable on small farms where low lift pumping is needed, where high value crops are grown and where the demand for irrigation is regular over much of the year. Solar pumping should be considered in future phases.,ofthese studies for water supply applications as well as for irrigation. Technical Development 0 0 0 0 Present field trial programmes should be continued wherever feasible The conclusions of this Report should be implemented with a view to producing improved PV systems. This may be done by specifying various improved systems with the desirable technical features so far identified, and ordering examples from suppliers capable of demonstrating a competence to respond to a call for tenders. The systems produced in this way should then be laboratory tested, prior to any further field testing, to confirm that they achieve specification and to identify any further shortcomings. Following satisfactory laboratory tests, such systems may subsequently be field tested. 180 0 Further design studies, supported by continued fieid testing of the more successful systems so far installed in the field should be carried out to investigate further possible areas of improvement and to improve the data base. 0 The development of small-scale thermal systems with concentrating and tracking collectors appears to be fruitful and should be encouraged. 0 0 5.3 A review should be made of any worthwhile improvements which may be worth considering to individual components of PV systems. Specification for manufacture should be prepared and tenders invited. Desk studies of the detailed requirements for local manufacture or assembly should be initiated to be followed up, if possible, by case studies. Institutional Arrangements Countries to be involved in field testing programmes should be able to satisfy a number of criteria, the most important being: 0 The existence of important pumping needs for irrigation and water supply in rural areas that could be met by solar powered pumping systems and which would require a range of pump output power suitable for solar systems. 0 The presence of a suitable solar energy resource and the absence of any more readily exploitable alternatives. 0 Government interest in solar pumping and a willingness and ability of host country institutions to provide the necessary technical and logistical support for the reliable field monitoring of the systems.. soume use Held W 11518 opm well N”nc 2.040 19J”le.O 3 25Sq,80 I hnptr C”i”d *wady in operation 3 SE, fSfU180 3 ,,,“,80 30 ,,h”RO 23 l33”“80 4 *osep*o 6 241”“80 32 26 1”,80 E fJ w 183 184 Factor Investigated Outline Description I. Static head variation investigate effect of changing static head from Z.Sm to iO.Om (all 9 reference systems). 2. Solar day variation Insolation data for a typical hazy day substituted for clear reference day data (all 9 systems). 3. PV array optimization Array nominal voltage varied keeping power constant to find optimum voltage. Voltage then held constant at optimum value and power varied to find optimum power (all 9 systems). 4. Impedance matching The use of a perfect array maximum power point tracker, first with zero losses and then with 10% of output power losses. Also crude impedance matching by series parallel switching af certain times of the day (three systems only). 5. Sun tracking The effect of continuous sun tracking my the array was compared to manual tracking with one, two or three array movements per day (at “correct” and at “incorrect” times of day), (three systems only). 6. Pipework variation The effect of changing system losses by increasing or decreasing delivery pipe length and diameter (two systems only). 7. Cost sensitivity The effect of changing the array cost relative to the balance of system N B. The effect of the changes investigated was assessedin terms of the daily pumped output and Specific Capital Cost. The mathematical model built as part of the systemdesignstudieswas used for this work. TABLE IV- SENSlTIVITY ANALYSIS OF PV PUMPINGSYSTEMS MODIFICATION Effecti&’ ) Specific Overall Capital efficiency Cost (%) (%I kJ per day) Daily output at 5m head (standard clear day (m3) Effect compared with basic system Volume Spec Cap pumped Cost None - System as supplied 2.2 3.4 44.9 I .oo 1.oo Voltage and power optimized by changing array cells series/parallel arrangement 2.6 3.0 46.6 (2) 0.88 MPPT ( 10% losses) (Maximum Power Point tracker) 2.5 3.1 51.2 1.14 0.91 Manual tracking of sun (2-adjustments of array position per day) 2.9 2.6 51.7 1.30 0.94 Voltage and power optimized plus manual tracking 3.3 2.3 60.3 (2) 0.68 MPPT plus manual sun tracking 3.1 2.5 62.3 1.39 0.74 Notes : (1) based on irradiation on fixed array. (2) power is reduced and so pumped volume is not comparable with basic system TABLE V - RESULTS OF MAKING ‘IMPROVEMENTS’ TO A PV PUMPING SYSTEM BY USING THE MATHEMATICAL SIMULATION MODEL 186 APPENDJX4 Objectivesof and Preparationfor PhaseII of the Project 187 APPENDIX OBJECTIVES OF AND PREPARATION 4 FOR PHASE II OF THE PROJECT* The next phaseof the Project, PhaseII preparation, providesa period in which to reflect on the resultsobtained from PhaseI, to confm the objectivesof PhaseII, and to make the necessarypreparationsfor it. The Consultantsthink it important to use this period to review the applications, economicsand systemsizesof solarpumps. As far asapplicationsare concerned, it will be necessaryto review the conditions under which solar pumps will be suitable for irrigation purposesand to evaluate in detail the potentially more attractive water supply application. The economiccriteria to be satisfiedby the pumpswill needto be examinedin more detail and the relative economicmerits of other lifting deviceswill need to be assessed, so that solarpumps are only demonstratedwhere there are good propectsfor their technical and economicviability. It will alsobe necessaryto study and defiie the pumping requirements(head, flow and pattern of consumption) in order to build up a profile of market requirements for eachapplication. The main areasof technical developmentshould be asfollows:0 Presentfield trial programmesshould be continued whereverfeasible. 0 The conclusionsof this Report should be implemented with a view to producing improved PV systems.This may be done by specifyingvarious improved systemswith the desirabletechnical‘featuresso far identified, and ordering examplesfrom suppliers capableof demonstrating a competenceto respondto a call for tenders. 0 The systemsproduced in this way should then be !aboratory tested, prior to any further field testing, to confirm that they achievespecification and to identify any further shortcomings. 0 Following satisfactory laboratory tests, such systemsmay subsequently be field tested. 0 Further designstudies,supported by continued field testing of the more successfulsystemsso far installed in the field, should be carriedout to investigatefurther possibleareascf improvement and to improve the data base. 0 The developmentof small-scalethermal systemswith concentratingand tracking collectors appearsto be fruitful and should be encouraged. 0 A review should be made of any worthwhile improvementswhich may be worth consideringto individual componentsof PV systems.Specification for manufacturemay then be preparedand tendersinvited. 0 Deskstudiesof the detailed requirementsfor local manufactureor assembly should be initiated to be followed up, if possible,by casestudies. * A copy of chapter 12 ‘Future Work’ from the Project Report 188 Visits needto be madeto potential PhaseII host countriesto explain the Project and explore the posaibllities for their future involvement. The countries to be involvedneedto satisfy a number of criteria the most important being: 0 0 0 The existance of important pumping needs for irrigation and water supply in rural areasthat could be met by solar-poweredpumping systems and which would require a range of pump output power suitable for solar systems. The presenceof a suitable solar energyresourceand the absenceof any more readily exploitable alternatives. Governmentinterest in solar pumping and a willingnessand ability of host country institutions to provide the necessarytechnical and logistical support for the reliable field monitoring of the systems. Additional visits will be made to selectfield trial sitesfor PhaseII and to agree and brief the participating institutions. The fmal objective of PhaseII ls seenasthe developmentof solarpumping systems to the stagethere they will be suitable for pilot manufacture or assemblyin developingcountries.
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project