Part 17 - cd3wd417.zip - Offline - SmallScale Solar Powered Irrigation Pumping Systems

Part 17 - cd3wd417.zip - Offline - SmallScale Solar Powered Irrigation Pumping Systems
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.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement