Canadian Water Resources Journal

ISSN: 0701-1784 (Print) 1918-1817 (Online) Journal homepage: https://www.tandfonline.com/loi/tcwr20

Applying Photovoltaic Technology to Pump Water

Eric J. Schiller & Brian G. Latham

To cite this article: Eric J. Schiller & Brian G. Latham (1988) Applying Photovoltaic Technology to
Pump Water , Canadian Water Resources Journal, 13:1, 52-61, DOI: 10.4296/cwrj1301052

To link to this article: https://doi.org/10.4296/cwrj1301052

Published online: 23 Jan 2013.

Submit your article to this journal

Article views: 232

View related articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=tcwr20
Applying Photovoltaic Technology to Pump Water
Eric J. Schillerl and Brian G. lathamz

Abstract.
In remote areas, particularly in developing countries, the use of photovoltaic (PV)cells to
provide electricity for water pumping is a technology which is being currently
developed. lt can be dependable and economical for small units and operates without
expensive energy supplies, Most systems couple a submersible centrifugal pump with
an AC or DC motor. Existing methods used to select units for a given location are com-
pared on the bases of the information required and the computations involved. Cunent
problems in operation and maintenance are discussed.

Resume.
Dans les 169ions isolees, plus particulierement dans les pays en voie de d6veloppe-
ment, on met actuellement au point une nouvelle technologie qui permettrait de pro-
duire de l'6lectncite pour le pompage de l'eau d l'aide de cellules photovoltdques. Cette
m6thode, qui s'av6rerait fiable et 6conomique pour les petites installations, ne requiert
aucune alimentation en 6nergre co0teuse. A I'heure actuelle, le pompage se fart au
moyen d'une pompe centrrfuge rmmergee, actionn6e par un moteur 6lectrique d
courant continu ou d courant alternatif. L'auteur compare les methodes utilisees pour
choisir les unitOs de pompage d'un site donn6, ll passe 6galement en revue les problOmes
d'exoloitation et d'entretien.

Introduction conditions which can persuade villagers to

The provision of safe water for drinking pur- remain in their rural home areas rather than to
poses in developing countries is a major con- emigrate to overpopulated urban areas. Provid-
cern for engineers, Besides the usual design ing adequate amounts of safe water in rural
concerns, systems often need to operate under areas is a vital prerequisite for development.
difficult constraints such as: However, to provide water by conventional
methods often requires large initial expenses to
1. low government budgets,
develop the infrastructure of the rural areas.
2, high imported fuel costs,
To overcome many of these problems, a
3, a lack of soare Darts to service motors water supply technology is needed that operates
and generators,
on its own local energy supply and that de-
4, lack of roads and electrical grids needed to
creases dependence on roads, electrical grids
operate and service water pumps.
and foreign fuel supplies. Such a technology is
The above problems are especially critical in rapidly coming of age and this is photovoltaic
remote rural areas, where it is very difiicult to (PV) pumping. While it is not suitable for all rural
provide energy and maintenance for rural water situations, future developments in the field will
supply schemes. To overcome these problems, make it more and more viable in many areas.
several new technologies are being developed.
Providing water In villages increases the time Description of Solar Pumps
available for productive work, particularly in A solar pump system consists of a water pump
agriculture, and reduces health problems, In- driven by an electric motor which derives its
directly, it has the effect of improving living energy from the sun (Figure 1). lt generally
consists ot: untuk peng
solar pump
1 Associate Professor, Civil Engineering Department, University of Ottawa, Ottawa, Ontario
2 Hydrologist, Government of Canada, Yellowknife, N.W.T.

52 Canadian Water Resources Journal / Yol. 13, No. 1, 1988

 FIGURE 1: schematic Diagram of a pV solar pumping system (Kenna and Glllett, 1985)

PV ARRAY

STORAGE
TAI{K
SUPPLY
PIPE

1, An array of photovoltaic cells. The number d) controls and switching electronics,

and configuration of the cells need to be e) DC-AC inverters for use with a AC motor.
determined for each application. These permit use of a wider range of mo-
2. A motor/pump unit. In most cases, a submer- tors, but they do use energy (.10 percent).
sible motor/pump system is employed with
photovoltaic pumping units. Although they increase efficiency, many of
3. Water storage facilities. An elevated storage these power maximizing features also make the
reservoir enables excess energy to be stored system more complicated and so they are often
for times of peak usage. omitted in practical field situations.
4. Power maximizing equipment including:
a) sun tracking sensors and motors, Advantages of PV Pumping
b) maximum power electronics (array mix- Solar pumps have a number of advantages
ers) to optimally select the combinatton when compared to diesel pumping. These
of current and voltage delivered by the include:
array that maximizes the power outout. 1. No external fuel source is required. In times
These require power (4-7 percent) but of uncertain fuel supplies, poor roads, strrkes
generally increase overall efficiency, and low foreign reserves, the pump con-
c) battery chargers and batteries, tinues to work.

Revue canadienne des ressources hydriques ..|,

/Vol, ir3, No. 19gg F,?
 2. The systems are very suitable for dry sunny Eskenazi el a/, (1986)) show that solar
climates, where the water source is under oumos are more cost effective than diesel
the ground. In fact the drier and sunnier the pumps for low power applications (about
climate, the more energy and, hence, the 1 kW). This is the same power range of many
more water can be delivered. of the small scale rural water needs in
3. Since PV pumping systems have "stand- developrng countries as well as small isolated
alone" capability, they are not dependent on applications in developed countries.
expensive expansions of the power grid. Properly installed, the units have an extremely
4. The match between centrifugal pumps'load long life (approximately 20-30 years for PV
line and the maximum power points of a PV panels) and the solar panels require a mini-
array is very good (Figure 2), Thus the PV- mum of maintenance. The PV cells in fact
centrifugal pump system can be made to have a lifetime much greater than the average
operate at near peak efficiency without con- motor/pump unit (5-10 years).
trol devices such as DC-AC inverters and The cost for the water pumped is steadily
power maximizers. This results in cost savings decreasing as the costs of solar cells de-
and greater reliability. crease and efficiency of cells, pumps and
5. Economic analyses (KennaandGillet(1985), motors increases.

FIGURE 2: Performance Characteristics of a Typical PV Pumping System, (Kenna and

Glllett, 1985)

I ooo w,/m 2, 55 0
c

8@ Wz62,59"g
TRANSPOSED
HYORAULIC LOAD
CHARACTERISTIC

600 w/mr, 45oC

F PV
ztrJ ARRAY
CHARACTERISTICS
t(r
f, 4oo w,hr,40"c
o LOCUS OF
MAXIMUM
POWER POINT
OF PV ARRAY

VOLTAGE

Details of PV Pump Components current when exposed to sunlight, A single cell,

Of the components mentioned above, those 6 cm diameter, usually develops about one watt
presently undergoing design improvements are of peak power. When these are assembled and
the PV cells and, to a lesser extent, the D.C. properly connected, any Eiven combination of
motorpump assembly. voltage and current can be produced. Units of
1-2 kW oeak oower are often assembled.
Photovoltaic Arrays A complete review of the types of PV cells,
A PV array is a series of cells, commonly made their efficiencies and their potential for use in PV
of thin layers of silicon that generate an electric pumping is presented in Table 1.

Canadian Water Resources Journal / Vol. 13. No. 1, 19BB

TABLE 1: Review of Solar Cell Efficiencest
(Sources: Bolton (1983), Bloss (1982) and others)

Type of Ce11 Comnercial

f - A r--,.
Array Power CommenEs
Convers lon
Efflclency Ef flclency
(Bo1 ton) ( Bloss )

Slngle-crys ta1 LZ-r4Z 12-r57. MosL connon for PV

sillcon pumping efflclency
r0-L3:l
S lngle-crys cal zzf/. Good future. Expenslve.
GaAs

CdS /Cu2S 3-sz 8-toz Potentlally lor{-cosc buE

Iow efflciency
New developmenEs coming,
noE connerclally
available, 5Z efflclency
Anorphous q!o/
6-77" Made nostly ln Japan
slllcon Future poEentlal ln PV
pumping
Not fully proven

Polycrys talline
cel 1s
-S I ll con LO-r2'l Future PV pump potenElal
-CdS /CuInSe2 8-to7" Not in cornmerclal
-GaAs 8.L7. r6-207. productlon for
CdTe LO.57. pump uses

Other new
materlals 0 .7 -L4 .8% Experimental

* Efficlency is the ratlo of elecErical energy ouEput to the solar energy

input ("sun Eo wlre" efficlency).

Photovoltaic arrays are specified by ratings of 3. The following factors can further reduce the
peak power output in watts, which indicates the efficiency of operation:
maximum power output at a celltemperature of (a) soiling of the array's glass by dust, dirt,
25"C and an irradiance level of 1000 W/mr. bird droppings,
These conditions are seldom present in the field (b) electrical losses outside of the array
and the actual array output will be much less itself,
than this for the following reasons, (c) operation of the array under conditions
1. Cell operating temperatures in tropical coun- other than its maximum oower ooint,
tries often rise to 50' or more which can Another factor that should be considered when
result in a 15 percent reduction in power. estimating operating life is that the efficiency of
2. Solar irradiance levels are more likely to the solar panel will decline due to degradation of
average 500 Wm, during the day. the cells by heat and solar radiation, This loss can
be about 5-1 0 percent over the first 5 years of use,

Revue canadienne des ressources hvdriques /Vol.13, No. 1, 19BB 55

 At the presenttime, PV cells cost about $6- 1O centrifugal pump system can be made to
US per peak watt or about $12-20 per effective operate at near peak efficiency without com-
or average watt. However, this trend is downward plicated control devices.
for the following reasons: 3. lt has a relatively low starting torque because
1, Markets for PV cells are just beginning to comparatively small rotor masses have to
develop. Mass production will lead to further
be accelerated, as compared to the case
with positive displacement pumps. Thus, no
economies of scale,
2. Research into the existing cells fabrrcated special starting mechanisms such as friction
from single-crystal silicon and newer com- clutches or centrifugal pulleys are required
pounds such as amorphous silicon and and the pumo and motor can be connected
gallium arsenide will result in more cost- directly together, As a result, the pump will
start at lower sunlight levels than positive dis-
efficient cells,
placement pumps.
Centrifugal Pumps 4, Under optimum conditions, the efficiency of
pumps can be up to 65-70 percent.
In photovoltaic pumping apptications, the most
suitable pump is the vertical axis, multi-stage, 5. The vertical axis allows the oumo to fit into a
centrifugal type. lt has a number of advantages
narrow well casing, as in most cases the
pump is a submerged type.
for photovoltaic pumping.
'1. lt can be used in low, medium
and high head Efficiency of Components
conditions although at very high heads, it is A number of efficiencies are quoted for the
less efficient than positive displacement various components of photovoltaic pumps. A
pumps. summary of the eff iciencies, quoted in the litera-
2. As noted above, the load line of a centrifugal ture, is in Table 2, Note that overall(sun to water)
pump is close to the locus of the maximum efficiencies are low, usuallv below 4 percent.
power points of a PV array. Thus, the PV-

TABLE 2: Summary of Solar Pumping Effaciencies

Component Efficiency (Z)

Sun Irradiance

7 5-90
PV CeIl Input
L2-L5 artay eff
PV Cell 0utput sun fo
85-90 motor eff notor/ wacer

Motor out.put puxnP wlre - z.o-o.u

70-40 pump eff comb. wacer:
Pump output 35-6 5 JU-.r )
70-90 hydraullc

eff
WaEer output

56 Canadian Water Resources Journal / Vol. 13, No. 1 , 1988

Suitable Sites capabilities of the pumping systems available
Theoretically, photovoltaic pumping can be will ensure that water is supplied at mintmum
carried out wherever the solar panels can be cost,
installed and groundwater is available. However, Manufacturers of particular systems will gen-
since the area of PV panels required to provide erally recommend the most suitable system for
power is quite large, the cost per cubic meter of a particular location. However, there are gener-
water pumped is relatively high compared to alized methods to select the most appropriate
surface water supplies. Therefore PV pumping system. Two typical methods are the ones pro-
is only economically feasible in specitic areas. vided by Grundfos and Intermediate Technology/
Suitable areas should have the following Sir John Halcrow Associates (1983), These two
characteristics: methods of selecting solar pumping systems
will be compared.
1 . There must be an abundant amount of solar A list of the parameters required f or the selec-
energy available, Daily sunlight intensity tion of PV pumping systems using the Grundlos
should be reasonably constant, Cloud cover and Halcrow - ITDG methods is given in
should be llmited throughout the year and Table 3.
there should be no extended periods without
sun.
Slte-System M atc hi n g M eth ods
2. The site should have a limited number of The two methods described above are quite
alternatives for the powering of pumps. In similar, Variations in the calculation of solar
general, this means that the area will not radiation and in the method of specifying the
have access to an electrical grid and a sup- system distinguish the methods. The Grundfos
ply of fossil fuels.
method is strictly applicable only to predeter-
3. The area should have very limited access to mined complete Grundfos systems because
surface water of suitable quality or quantity. these are the only systems for which charac-
4. Underground water supplies should be avail- teristic curves are provided. However, if the
able at a reasonable depth (less than 30-40 oroper curves were available for other com-
m for present operating systems), plete systems, this method would be applic-
While these requirements may seem to restrict able, The ITDG method is more general lt
the application of PV pumping technology to attemots to provide information to select the
very small areas of iire wc,rici, they do in fact de- individual components of a system. However, in
scribe remote rural areas in many developing practice, it may be too simplistic in the way that
countries. The senior author of this paper has components are matched, Some steps are
direct experience in the very large Sahel area omitted, such as the method to specify the con-
south of the Sahara desert where PV pump figuration of the cells in an array so that they give
technology has been applied to the local an optimum power match with the motor and
conditions. pump, Only the peak wattage rating of the array
Photovoltaic pumping has become cost is specified. Overall system curves are not pro-
competitive with diesel pumps in certarn areas. vided. These system curves generally need to
Solely on a cost basis, Kenna and Gillett (1985) be experimentally determined for a given site
estimate that this point is reached when the Thus no overall rating of efficiency can be
average daily hydraulic energy requirement derived without field tests
(daily volume pumped x head) is less than 250 For situations where a test site is available, it
m4 and average daily radiation is greater than is possible to obtain the operating characteris-
10 MJ/m2, Eskenazi er a/, (1986) consider that tic curves for a specific pumping system from
with less conservative, yet realistic assumptions, Dump tests. One can then use these curves
PV solar pumping is clearly the least cost solu- directly to select the pump for a specific site.
tion up to a daily hydraulic energy requtrement Tests can be conducted to obtain the following
of 625 m4, and rn certain situations can be cost instantaneous data: Solar radiation (W/m'?), Dis-
competitive for values of hydraulic energy far in charge flow (m3/s), Total pressure head (m). lt is
excess of this. then possible to plot these data, This was done
for a TPK pumping system equipped with a
Selecting the Correct Unit for the Site Jacuzi immersion pump ln Dakar, Senegal and
As is the case for other water supply systems, the results are plotted in Figure 3. With such curves'
PV pumping systems must be chosen to suit the if the daily solar radiation is known, it is possible
specific location, The requirements of each site to calculate the amount of water pumped directly
are different and the matchinq of these to the by integrating the water pumped over discrete

Revue canadienne des ressources hvdriques /Yol 13, No 1, 1988 57

TABLE 3: Information Required for PV Pump Selection
Note: * : site specified parameter

INFORI'IATION REQUIRED INFORMATION SUPPLIED

irDG
-GrunaTos

L. Slte IocaEion
z. SiEe Iat.lEude

Average daily waEer demand (monEhly)

Hydraullc head
- tocal *
- plpe headloss estlnate Implied only
Average daily radlation (monthly)
a) directly fron tables & maps
b) from monlhly combinaEion of
- maps of clearness index
- Eable of average datly global
exEra-EerresC.rial radiation
6, Tilt fact.or, function of:
- Iatltude, month, EIIE angle Graphs
- latltude, month, EiIt angle,
clearness index
7. System Infornatlon
a) Conplete sysLem characterlsEics
- Vol. pumped vs. hydraullc
head for one radlaElon level Graph
- Vo1, punped vs, radiaclon
and hydraullc head Graphs
b) Conponent. deslgn lnformation
Mntnr/nrrnrn
- rrvLv!, Puu't, "nit
u. efflclencies
- typical overall Table
- peak power TabIe
- Pump Lypes vs daily volume
and hydraulic head Graph
- Avallable plpe slzes *
- llomographs, deslgn aids
- artay peak waEts vs.
hydraulic energy, pump/motor
efflclency, solar radiaElon Graph
- peak flow race vs. motor/PumP
peak efficlency, head, array rating Graph
- pipe head loss vs. flow rate,
pipe diameter Graph

8. Cosl of other options to supply

the same amount. of useable waEer Jr

5B Canadian Water Resources Journal / Vol. 13, No 1' 1988

 FIGURE 3: Head-Discharge Curves as a Function of Sunshine (Schiller, 1986)

eo.oo

72.OO

ul
I

F
o
F

r20.oo t60.oo 200.oo 240.OO 280.OO 320.OO

FLOW (L/Hl a tO

time intervals during the day, The average solar By using erther manufacturers' curves, or
radiation for a period of time could also be used experimentally derived curves from the field it is
to compute the average volume ol water pumped possible to accurately predict the amount of
per day during the period under study. A com- water that can be obtained using a solar
parison of typical observed and calculated powered pumping system, if the solar radiation
results for Dakar, Senegal (Schiller, 1986) is and the depth to a reliable water table are
given rn Table 4. Known.

Revue canadienne des ressources hvdriques /Yol' 13, No 1, 19BB 59

TABLE 4r Comparlson of Water Pumped. Dakar, Senegal

OBSERVATIONS CALCULATED DIFFERENCE

Tlme Flow Volume

w/n2) (L/h) (m3)

l3-r4h L ZIlU 1.24

14-l5h 1200 L.20

15-l6h 900 0.9

Operation and Maintenance are commonly used in PV pumping in order to

At present, a body of experience in operating avoid the use of an inverter required for AC
solar pumping systems as accumulating (Sy and pumps, However, these motors require a cas-
Gia, 1984; Billerey, 1984). Costs of operation ing that must be carefully sealed. The motors
and maintenance (O/M) represent the only cost themselves have experienced brush wear prob-
after the initial capital outlay of the unit. A recent lems. However, manufacturers are presently
study in Mali (Billerey, 1984) has registered an working on these problems and improvements
O/M cost of about US$.08 per m3 of water are being made,
delivered. The experience with these systems in Finally, some photovoltaic pumping systems
the field indicates that the ohotovoltaic oanels use above ground motors connected to the
are the most reliable component in the system, pump by a drive shaft. These units require
In dry, dusty areas they do require regular regular electrical and mechanical repair.
cleaning, but this can be easily done, often with
a cloth. Occasionally, panels break, but this is Future Prospects
often due to carelessness or vandalism, A At present, for systems greater than 1-2 kW, the
secure fence, together with community sur- cost of the photovoltaic panels are the major
veillance can usually secure the panels, Finally, factor in the overall cost amounting to about 60
there is a long term degradation of the panel percent of the total system cost for larger sys-
efficiency due to the combination of heat and tems (Figure 4).
solar radiation. This has been measured at 5- 1 0 There has been a dramatic drop in the cost of
percent over an initial period of five years. PV cells in recent years. (World Water, 1983;
The greatest number of operational failures Hamakawa, 1987). The prediction is that these
have been recorded with (1) the borehole that costs will continue to drop in the future as a
either becomes dry or silts up due to inade- result of technological progress and improved
quate welldevelopment and (2)the submersible methods of mass production, lf these trends
pump units, These are not new problems caused continue, and if the maintenance problems can
by the photovoltaic pumping systems but are be resolved, then solar systems may become
common to other pumping systems. Therefore, the dominant water pumping technology within
standard O/M procedures need to be carefully the next decade in many sunny, arid and rural
and consistently applied. DC submersible pumps regions in developing countries,

60 Canadian Water Resources Journal / Vol. 13. No, 1, 1988

FIGURE 4: Costs of Solar Pumping Systems in Mali (Billerey, 1984)
COST/WATT
FRENCH FRANCS

L CONSTRUCTION

CIVIL ENGINEERING AND MOUNTING COSTS

PV PANELS ANO' / /
MOTOR-PUMP SYSTEMS

to 15
PEAKKfOWER

References
Billerey, J. 1984. Le Pompage Photovolhique. Norman, C.E D.G. MacKenzie and C.J. Robins,
GRET/GERES/AFME, 17 rue Xantrailles, Paris. 1984. "Photovoltaic Water Pump Develop-
112 p. ment", TPK Solar Systems Inc,, Ottawa. 65 p.
Bloss, W.H. 1982. "Photovoltaic Energy Conver Norman, C.E., C, Robins and B,G. Keefer, 1984.
ters" in Performance of Solar Energy Conver- "Final Report on Development
ters: Thermal Collectors and Photovoltaic Cells. of Photovoltaic Powered ShallowWell Pumping
Edited by G. Beghi, R. Reidel Publrshing, Dor- System", TPK Solar Systems Incorporated,
drecht, Netherlands. pp. 147-162. Ottawa, 130 p.
Bolton, J.R. 1983. "Solar Cells - A Technology Schiller, E,J. 1986, "Fapport sur les essais
Assessment", Solar Energy, 31(5), pp 483- techniques du systeme de pompage solaire
502. TPK-D". CERER, Dakar, S6negal. 42 p.
"Engineering Design Manual for Grundfos Solar Sir William Halcrow and Associates and Partners.
Pumping Systems". Internal Manual, Undated. 1983. Sma//-Scale Solar-Powered Pumptng
Systems: The Technology, its Economtcs and
Eskenazi, D., D. Kerner and L, Slominski. 1986,
Advancement Main Report. Intermediate Tech-
"Evaluation of International Photovoltaic Pro-
nology Power Limited, London. 3OO p,
jects". Sandia National Laboratories, California.
360 p. "Solar Breakthrough for Small Sahel Systems".
1983. World Water. September, pp. 35-36.
Hamakawa, Y, 1987. "Photovoltaic Power".
Scientific Amerrcan. April, pp. 87-92. Sy, B.S. and L.H. Gia. 1984, "Etude-Diagnostic
sur les Pompes Photovoltarques et Eoliennes
Kenna, J, and W. Willett. 1985. Solar Water
au 56negal". CERER, Universit6 de Dakar,
Pumping; A Handbook. lntermediate Technol-
Dakar,S6n6gal.11p,
ogy Publications, London, 1983. 123 p.

Revue canadienne des ressources hvdrrques /Yol.13, No. 1, 19BB t)l