4.  POWER FOR PUMPING

Fig. 81 indicates the most feasible linkages between different energy resources and prime movers. It shows how all energy sources of relevance to small or medium scale irrigation pumping originate from renewable energy resources or from fossil fuels; the arrows then show all the routes that can apply from an energy resource to produce pumped water. In some cases similar components can be used within systems energized in completely different ways; for example electric motors are necessary either with a solar photovoltaic pumping system or with a mains electrical system, so the motor-pump sub-systems of both types of system can have a lot in common.

The details of the components in Fig. 81 are discussed through the following section, but it is first worth reviewing a few generalities relating to the combination of prime movers and pumps.

4.1    PRIME MOVERS AS PART OF A PUMPING SYSTEM

4.1.1 Importance of "Cost-Effectiveness"

Almost every aspect of an irrigation pumping system consists of compromises, or trade-offs, between the capital (or first) cost of the system and the running (or recurrent) costs. Farmers tend to purchase cheaper systems with higher running costs, resulting in the widespread use of other than the most efficient and cost-effective systems. No doubt this is because farmers are generally short of capital and often in any case regard large capital investments as being inherently more risky than incurring regular running costs, (which in time may mount up to a large sum).

There is therefore a good case for institutional users with an interest in improving agricultural techniques, such as agricultural credit agencies, aid agencies and governments, to try to assist in overcoming these problems by providing suitable financial inducements to encourage the use of more cost-effective irrigation systems.

In the final analysis, the combination of components to make a pumping  system will depend on the cost effectiveness of the total chosen system under whatever specific technical, agricultural and financial conditions happen to be prevalent. That is "cost-effectiveness" in the very broadest sense, including not just the first costs and running costs for the system, but factors like the convenience and ease of use as perceived by the farmer; (i.e. including, in economists' jargon, the "opportunity costs"). In reality, the selection decision is usually limited to what is known to be available and affordable and yet is capable of the required pumping duty.

The selection process is discussed in more detail in the next section, but certain considerations inherent in the economics and hence in the relative cost-effectiveness of different system choices are important in relation to any discussion of linking components to make up pumping systems. To this end, some of the cost attributes of different prime-mover options are compared in Table 9 . Here the main categories of irrigation pumping system are reviewed in terms of first costs, recurrent costs and then in terms of two factors relating more to their "effectiveness" (i.e. their productivity and their general availability). It is clear from this table that no single method offers both low first costs and low recurrent costs and yet is also among the most productive (otherwise it would be universally applicable and the other options would be of little interest). High productivity depends either on relatively high first costs (when using renewable energy equipment generally) or on relatively high running costs combined with moderately high first costs when using fossil-fuelled equipment. In the end, successful selection depends on the choice of the best trade-off between the availability of finance, the capability of maintaining and financing the recurrent costs of the system and the performance or productivity that is expected.

Fig. 81 Linkages between energy resources and appropriate prime movers

Table 9   COST ATTRIBUTES OF PRIME MOVERS

¹ Electricity can vary from low to high depending on whether a mains connection is already available or not.

² "Attendance" implies level of human intervention needed.

It is possible that potentially useful systems are usually not considered simply because, not being conventionally used at present, they are unfamiliar and therefore are not known or understood well enough by potential buyers; this is where it is hoped that publications such as this may encourage some attempts to try new methods, preferably by institutions or individuals with the resources to underwrite the risks inherent in experimenting with new or unfamiliar technologies.

i. Cost-effectiveness and efficiency

Generally, a cost-effective system needs to be technically efficient; i.e. a relatively high output is needed in relation to the energy input. This is just as true for renewable energy powered systems  as  for fossil  fuelled  systems. In the former case, the energy resource, if it is solar energy, wind or water power is notionally cost-free, but the capital cost of the system is closely linked to the efficiency. This is because for a given  pumping requirement, if you halve the efficiency of the system you must double the "cross section" of the energy resource to be intercepted; i.e. you need twice the area of solar collector or twice the rotor area of a windmill, or a turbine capable of passing twice the flow rate of water. This tends to require a system that is twice as large and therefore usually twice as expensive.

In all cases there is an ultimate technical efficiency that can be approached but never quite achieved, (Fig. 82 A). Pursuing the cause of better efficiency is usually worthwhile up to a point, but thereafter it brings diminishing returns as increasing complication, sophistication and cost is required to achieve small further gains in efficiency. However it usually requires a mature technology to be at the level where further improvements in efficiency are counter-productive, and in any case, new manufacturing processes and materials or increases in recurrent costs (due to inflation) sometimes allow improvements to become cost-effective in the future which were not justifiable in the past.

The influence of efficiency on costs is illustrated in Fig. 82 B, which shows how low efficiency generally causes high costs and that there is an optimum range of efficiency for most technologies where reasonably low costs are achieved, but above which diminishing returns set in. In the case of renewable energy systems these costs will be largely attributable to the capital cost and hence to financing the investment, while in the case of fossil fuelled devices a large proportion of the costs will relate to running and maintaining the system.

Fig. 82 A. How diminishing returns eventually defeat the benefits of seeking increased efficiency beyond certain levels 
B. The influence of efficiency on costs

ii. Combining system components with differing efficiencies

Virtually all pumping system components achieve an optimum efficiency at a certain speed of operation. Some components like pipes and transmission systems are most efficient (in terms of minimizing friction and hence losses) at very low rates of throughput, but they are then least productive and they will therefore have a point of "optimum cost-effectiveness" where there is a good compromise between their productivity and their efficiency. Prime movers invariably have an optimum speed of operation; this is as true of humans and animals as it is of diesel engines or windmills.

Fig. 83 shows three sets of curves; first, efficiency against speed for two prime-movers, (in this example electric motors would fit the speeds and efficiencies shown); second,a curve for a typical pump and lastly, for the combination of the prime-movers with the pump. It must be remembered that the efficiency of a combination of two components is numerically the product (i.e. the multiplied result) of their individual efficiencies eg. a 30% efficient engine (0.3) with a 50% efficient pump (0.5) has a combined efficiency of:

Fig. 83 Illustration of how correct speedmatching of a prime mover to a  pump can be more important than the efficiency of the prime mover. Here prime mover 'A' is less efficient than 'B' but has a better speed match to the pump — hence the less efficient prime mover 'A' provides a more efficient system

The important point contrived in Fig. 83 is that the prime-mover with the highest optimum efficiency is not in this case the best one to use with a particular pump. In the example motor "A" has a best efficiency of 58% while motor "B" achieves 66%, yet, because the optimum efficiency of motor "A" occurs at a speed which coincides well with the optimum efficiency of the pump, the combined efficiency of that combination is better if the motor is direct-coupled to the pump; (motor "B" will drive the pump at a speed greater than its optimum, as at 1 500 rpm the pump has an efficiency of only 35%) so the best efficiencies of the two alternative combinations are:

This illustrates how it is generally more important to ensure that the design speeds of components match properly than to ensure that each component has the highest possible peak efficiency.

4.1.2    Transmission Systems

Components often do not match effectively; i.e. their optimum speeds of operation are different. In such situations it generally pays, and it is sometimes essential, to introduce a speed changing transmission. Also, in many situations the prime-mover cannot readily be close to the pump, and some method is therefore necessary for transmitting its output either horizontally or vertically to the water lifting device.

i. Transmission principles

Power can be transmitted from a prime mover to a pump in a number of ways; the most common is a mechanical connection, which can either rotate (shafts, belts or gears) or reciprocate (pump rods or levers). Where power has to be transmitted some distance, then electricity, hydraulic pressure or compressed air can be used, since it is difficult to transmit mechanical power any distance, especially if changes of direction or bends are needed.

In all transmissions there is a trade-off between the force or torque being transmitted by the system (which demands robustness to resist it) and the speed of operation (which tends to cause wear and reduced life). Power, which is what is being transmitted, can be defined as the product of force and velocity. Mechanical systems that run at slow rotational or reciprocating speeds need larger forces to transmit a given amount of power, which in turn require large gear teeth, large belts or large pump rods (for example) and these inevitably cost more than smaller equivalents. Where mechanical power is transmitted some distance any reciprocating linkages need to be securely anchored; (even a 5m farm windpump can pull with a reciprocating force peaking at about 1 tonne). For this reason, most modern commercial systems involving lengthy mechanical links tend to use high speed drive shafts (for example surface mounted electric motors driving a rotodynamic pump located below the water (or below flood level) as in Figs. 66 or 134 B). A high speed drive shaft can be quite small in section because its high speed results in low torque. However a high speed drive needs to be built with some precision and to have good (and expensive) bearings to carry it and to align it accurately so as to prevent vibrations, whirling of the shaft, premature wear and other such problems.

Electrical, hydraulic or pneumatic transmissions all have a common requirement demanding that their voltage, or pressure of operation ideally  needs to be high to minimize the cross section of cable or of pipe needed to transmit a given power flow efficiently. High voltage cable (or high pressure pipes), need to be of a good quality and inevitably cost more per metre for a given cross section. Therefore, with all transmissions there is a trade-off between efficiency and cost; cheap transmissions often reduce the capital costs but result in high recurrent costs due to their lower efficiency and greater maintenance and replacement needs, and vice-versa. It is therefore advantageous to match prime-movers and pumps of similar speeds to avoid the  cost and complication of speed-changing transmissions.

ii. Mechanical transmissions

 The most common need for a mechanical transmission is to link an  engine or an electric motor with a pump. Generally such prime-movers are used with centrifugal or other rotodynamic pumps which run at the same speed as the engine or motor; in such situations they can be direct-coupled with a simple flexible drive coupling as in Figs. 72 and 105. Speed changing of up to about 4:1 can readily be achieved with vee-belts as shown in Figs. 84, 99 and Fig. 106. Fig. 84 shows a two stage vee-belt drive where the total speed change can be as much as 4:1 on each stage. In this situation the total speed change is the product of the ratios for each stage. Where multiple vee belts are needed on one drive stage, as in Fig. 99 (showing four in use), it is best to use matched sets from a supplier, and always to renew all belts simultaneously so that they all share the load effectively; a more modern and convenient type of belt is the so-called poly-vee, which is similar to a whole lot of small vee-belts fixed together edge-to-edge. Flat belts (made of leather) used to be common and they are coming back, sometimes today made of synthetic materials, as they are more efficient with less friction than a set of vee belts.

Fig. 84 Two stage speed reduction transmisson used in China to connect an electric motor to a chain and washer pump

If a speed change greater than about 4 or 5 to 1 is needed, then an  alternative to multiple stages of belts (which introduce problems with belt adjustment) is to use gearboxes. A right angle drive may be created to drive a vertical shaft borehole pump (for example) either by using a 90° geared well head, or by using a twisted flat belt. To be successful twisted belt drives need to have a generous distance between the pulleys in relation to their diameters or excessive wear will occur.

 Other mechanical transmissions commonly used are reduction gearboxes  with a pitman drive, similar in most respects to the windpump transmission of Fig. 109 They consist of a rotary drive shaft which drives a single or pair of larger gear wheels via a small pinion; the large gear wheels drive a reciprocating cross-head or pitman slider via two connecting rods. The pump rod is connected to the cross-head or pitman. Mechanisms of this kind can be used to connect a diesel engine or an electric motor to a reciprocating piston pump. Other mechanical right-angle drives are illustrated by reference to Figs. 94, 95 and 96, (the large size necessary for making a strong enough drive from traditional materials is well-illustrated in Fig. 95).

When budgeting for a pumping system, it is important to know that the mechanical transmission can cost as much or often more than the prime-mover,  especially if a geared or reciprocating well-head is used. The high cost is due to the mechanical requirements for reliable operation being demanding and the volume of production usually being much lower than for engines or electric motors.

An effective method of transmitting mechanical shaft power any  distance is via a high speed rotating shaft. This needs to be steadied by bearings at quite close intervals to prevent the shaft "whirling" like a skipping rope, a phenomenon which causes intense vibration and destruction of the shaft. Vertical drive shafts down boreholes, as much as 100m deep and running at 1 500rpm or more are commonly and successfully used, although submersible electric multi-stage pumps are becoming a more popular solution.

iii. Electrical, hydraulic or pneumatic transmission

The use of a diesel-generating set (or wind-electric, solar-electric or hydro-electric unit) as a prime mover allows considerable flexibility in  transmission (literally) since electric cable is all that is needed to link the prime mover to a motor-pump unit, (which can even be submerged down a borehole as in Figs. 71 or 134 A).

 Other options, which are technically feasible, but more rarely used are hydraulic or pneumatic transmissions in which either a liquid (water or oil) or air are pumped through pipes to drive a pump. Examples of hydraulic  transmissions are given with the jet pump in Fig. 75, or the positive displacement hydraulically activated pumps of Fig. 39. The air lift pump of Fig. 76 is an example of a pumping system which requires pneumatic transmission. Pneumatic diaphragm pumps are commercially available and tend to be most commonly used for construction projects, with an air supply from mobile engine driven air-compressors. They are not normally used for on-farm irrigation but there is no technical reason why they would be unsuitable. However, hydraulic and pneumatic transmissions tend to be inefficient and therefore such a system may have high running costs.

4.1.3 Fuels and Energy Storage

Power sources need energy, whether it is fuel for an engine, wind for  a windpump or sunshine for a solar pump. The main difference is that the provision of fuel can usually be arranged by the user, but nobody can make the wind blow or make the sun shine on demand. There is therefore an obvious qualitative difference between wind and solar powered devices which will only function under certain weather conditions and the rest which generally can be made to operate at any pre-planned time.

Although the apparent randomness of wind or solar availability would appear to be a serious disadvantage, in reality the energy available over a  period of a few days in a given location at a given time of the year does not vary much from year to year. The problem is more one of covering a mismatch  that can occur between the rate at which energy is available and the duty cycle the farmer would like to impose. This can often be overcome either by choice of technique or by including a storage facility.

 In most cases, where the output required is water, the most  cost-effective solution is to introduce a storage tank between the pump and the field; (in some cases the field itself can act as a storage tank). The other principal method for small scale energy storage is to use lead-acid electrical batteries, but this becomes prohibitively expensive except when small amounts of energy of less than about l-2kWh need to be stored. The costs of tanks for storing water relate to their volume, while the costs of batteries relate to their energy capacity; therefore, at low heads when large volumes of water may need to be stored, but which involve little energy, electrical battery storage can be cheaper (and less demanding in terms of land utilisation) than storage tanks. However, before considering the substitution of batteries for storage tanks, it must be remembered that batteries would need replacing a lot more often than the storage tank and would also need much more maintenance.

4.2    HUMAN POWER

4.2.1 Human Beings as Power Sources

In the whole small-scale pumping field it is generally difficult to make precise statements on pumping performance which are generally correct; nowhere is this more true than in the field of human powered waterlifting devices and pumps. This is partly because human capabilities are very variable, but also because there is a multiplicity of pumps and water lifts of widely varying efficiency.

i. Efficiency as prime movers

People (and animals) derive their power from the calorific content of  their food. Even when physically inactive the human body requires energy to run its basic metabolic functions, i.e. to power the heart and circulate blood, to work the lungs and digestive system, etc. Energy for muscle power is then an extra requirement on top of this. A typical food energy requirement is around 2400kcal, 10MJ or 2.8kWh per 24 hrs. Table 10 indicates the calorific values of various staple foods (after Leech [6]).

A person's muscular work capability per day is  in the region of 200-300Wh/day. Human beings therefore have an average overall efficiency in the region of 7-11% for converting food energy to mechanical "shaft energy". This figure includes the basic metabolic energy requirement; the efficiency of the muscles for short but strenuous efforts can be as high as 20 or 30% [12] & [19], which compares well with internal combustion engines.

Table 10 THE CALORIFIC VALUES OF VARIOUS STAPLE FOODS

After Leech, reference [6].

Kraatz, [12] (quoting Wood), gives a calculation of the food required by a man to generate the energy needed to irrigate a crop. On the assumptions of a rice crop needing 850mm of water in 120 days, with a yield of 600kg of rice from a 0.2ha plot, with a 50% efficient water lifting device lifting the necessary water through 3m head, the marginal cost of "fuelling" the human prime-mover for the irrigation pump was calculated to be 35kg of rice or 6% of the expected total yield. An additional 35kg would be needed to cover the basic metabolism of the person concerned, giving a total of 70kg or 12% of the rice produced.

The Intermediate Technology Development Group's Water Panel [20] gave a rule of thumb of a food requirement of 0.5kg of rice per MJ of hydraulic work, plus 0.012kg of rice per day per kg body weight. In the earlier example, the hydraulic requirement was 50MJ, which under the rule of thumb just quoted demands 25kg of rice for pumping effort, and a 60kg man would additionally need (0.012 x 60 x 120 = 86kg) of rice, giving a total requirement of 111kg of rice, or 18% of the total crop produced.

Allowing for losses of rice, possible worse yields than that assumed, and the food requirements of the farmer's dependants, it is easy to see how hard it is to generate a surplus when cultivating staple crops on small land holdings. For example, if he has three dependants and loses just 20% of his crop through various forms of wastage, the farmer and his family will need to retain 60 to 90% of the harvest, depending on the method of estimating rice requirements used. Slightly worse wastage or a larger family would result in barely sufficient food for pure subsistance.

ii.     Productivity

Contrary to popular belief, human muscular energy is not cheap. The poor are forced to use human power, usually because they cannot afford anything better, since the cash investment required is minimized and therefore it is more "affordable" than other options. As will be shown, almost any other source of power will pump water more cheaply unless only very small quantities are required.

 The human work capability is around 250Wh/day, so it  takes four days' of hard labour to deliver only one kWh - which a small  engine could deliver in less than one hour while burning less than one litre of petroleum fuel. So the farmer with a small mechanized pumping system has the equivalent of a gang of 20 to 40 men who will work for a "wage" or running cost equivalent to say 1 litre of fuel per hour; not surprisingly, any farmer who can afford it will sooner choose to employ an engine rather than 20 to 40 men. This argument can be turned on its head to show the high price of human muscle power, if the "opportunity cost" or a real wage cost is assigned to human muscular labour; eg. assuming a daily wage rate of US $1.00/day gives an energy cost of about $4.00/kWh. Although this is a low wage for hard labour, even in some of the poorer countries, it represents an energy cost that is significantly more expensive than even new and exotic power sources such as solar photovoltaic panels.

 There is an opportunity cost caused by diverting people from more  important work to pumping water; the best asset people have is brains rather than muscle; therefore, if agricultural productivity is to improve and economic standards are to be advanced, it is essential to introduce more productive power sources for all except the very smallest of land-holdings.

iii.    Power capability

 Muscle power can handle quite large "overloads" for short periods, but  the power capability diminishes if more than a few minutes of activity are required. The power availability is also a function of the build, age, state of health and weight of an individual; finally the ability to produce power depends on the nature of the device being worked and the muscles that can readily be utilized. Table 11 A indicates power outputs that may be expected for individuals of 20, 35 and 60 years age respectively over periods of operation ranging from 5 minutes to 3 hours, after Hofkes [21], presumably with devices allowing much of the body to operate. Table 11 B shows actual results measured by the Blair Research Institute in Zimbabwe, [22].

 Therefore, although the actual output from any human powered pump is  not precisely predictable, an approximate prediction can reasonably be made. Fig. 85 gives a set of curves indicating the capability of from one to four people each providing 240Wh per day of useful work through a pumping device with an efficiency of 60%. These curves probably represent what is generally achievable under favourable circumstances and indicate, for example, that the daily output per person if lifting water 5m, is about 12m³.

Table 11 POWER CAPABILITY OF HUMAN BEINGS

PUMP 1.   Pumping head 10.54m 
time to fill 20 litre can           

PUMP 2.   Pumping head 6.35m 
Time to fill 20 litre can 

 PUMP 3.   Pumping head 2.14m  time to fill 20 litre can

Fig. 85 The number of people required to provide a specified quantity of water at different lifts. The curves have been derived by assuming that a single person can provide 60 watts of power for 4 hours per day and that the efficiency of the pump is 60%

iv. Ergonomics

The actual useful output from a person depends a lot on the way the water lift or pump works; the most powerful muscles are the leg and back muscles while the arm muscles are relatively weak, so conventional hand pumps are less effective at "extracting work" from a person then a device like a bicycle. Moreover, the "ergonomics" of the design are important; the operator needs to be comfortable and not contorted into some difficult position, so the device should require a relaxed posture, with the user well-balanced, and it should function best at a comfortable speed of operation. Utilization of the leg muscles will also often allow the operator to throw his or her weight behind the effort in order to gain further pedal pressure. Wilson [23] reported that a rotary hand pump was improved in output by a factor of three, (300%), by converting it from hand operation to foot operation. The same article also promoted the bicycle as a supreme example of effective ergonomics; it uses the right muscles in the right motion at the right speed and applies human power through a light but strong and efficient mechanism. Wilson makes the very valid point that what is needed is a pump which is as well designed, strong, efficient and easy to use as the bicycle. He quotes dynamometer tests as indicating that the average cyclist works at 75W when cycling at 18km/h; if this output could be produced while pumping, the following flowrates should be realisable at various lifts, assuming a water lifting device of only 50% efficiency:

For this reason the most effective irrigation pumps are in fact foot operated. Also an irrigation pump requires to be operated perhaps for several hours and therefore efficiency and ease of use are crucial. Hand operated devices are easier to install and can be lighter and smaller (since no one has to stand or sit on them and the forces that can be applied will not be so great anyway). Where pumps are used for water supply duties rather than irrigation, efficiency is less of a stringent requirement since any individual user will generally only operate the pump for a few minutes per day to fill a few small containers.

Therefore the criteria for defining a good human powered irrigation pump are significantly different for those for a water supply pump and it may be a mistake to use pumps for irrigation duties that have only been proved successful in water supply.

4.2.2 Traditional Water Lifting Devices

Many of traditional water lifting devices are particularly designed for low lift irrigation, and they are often foot-operated since it no doubt became apparent that this was the best method of harnessing human power.

The least-cost solution has always been a bucket or bag of water lifted when necessary on a rope; Fig. 86. The best that can be said for this technique is that with small plots the water can at least be applied with precision to individual plants, so at least efficient conveyance and distribution can partially compensate for the inefficiency of the actual water lifting. At low heads, the use of buckets and scoops, (see also section 3.3.1) led to the development of the swing-basket, (Fig. 18), which can use two people and functions more rapidly, although only through very low pumping heads such as from canals into paddies. However, it is not an ergonomic device in that a lot of muscular effort goes into twisting the body, there is much spillage, and also water is lifted much higher than necessary. Nevertheless, two young boys using this technique, for example in Bangladesh, can complete 2 000 swings without a rest, according to Schioler [91].

Fig. 86 Rope and bag water lift from a dug well (example from The Gambia)

Fig. 87 Relative performance of the swing basket and the Dhone (after Khan [25])

  An improvement, obtained at the price of some slight complexity, is  the use of suspended or pivoted devices such as the supported scoop (Fig. 19) and some which are also balanced such as the dhone (or dhoon) see-sawing gutters, (Fig. 20) or the counterpoise lift (or shadoof), (Fig. 21). These are no longer portable since they need to be installed on a site, and they require a supporting structure which has to be attached securely to the ground, but they are far more efficient than such primitive devices such as buckets or swing baskets, as indicated by the performance curves taken from Khan, [25], in Fig. 87. This shows that a single dhone will lift 7.5 litre/sec at a lift of 0.75m (115 gall/min at 30 inches), which reduces to about 2 litre/sec (30 gall/min) at 1.5m (or 60 inches) head. Therefore the dhone will move more than twice as much water as a swing basket at low lifts, moreover using the power of only one person rather than two. Khan makes the point that many Bangladesh farmers try and use a single stage dhone at too high a lift, and lose a lot of performance as a result; the optimum lift per stage is approximately lm (40in).

Table 12, adapted from Khan, [25], indicates how widely the dhone is  used in Bangladesh, a country with very large areas offering the possibility of shallow lift irrigation, and it also shows clearly how much of an improvement the dhone is over the swingbasket. The same table also indicates the characteristics relating to the "dugwell", a counterpoise lift and shallow hand-dug well, and the "Mosti", which is a cast iron handpump mounted on a tubewell.

 Despite being a considerable improvement on the swingbasket, the dhone  is probably not as efficient as the various rotary devices, described in Section 3.6, although it represents a good compromise between retaining simplicity and low-cost while achieving a useful output.

Table 12 COMPARISON OF VARIOUS WATER LIFTS IN BANGLADESH¹

¹Adapted from Khan, [25] 

²Hand dug well with counterpoise bucket lift 

³Manually operated shallow tubewell for irrigation 
Uses industrially made headpump (No. 6. Carbiron pump)

Rotary devices tend to be easier to work as they generate a smooth output and they therefore often can be driven with a comfortable pedaling motion of either the arm or, better, the legs. The various flash-wheels and  ladder or dragon spine pumps are generally leg-powered, (see Sections 3.6.5 to 3.6.7 plus Figs. 51 and 52), and they can often readily be powered in this way by several operators. While today Archimedean screws are usually hand operated (Fig. 48), in Roman times they were walked on rather like the treadwheel in Fig. 51, which no doubt was easier for the operator and more productive.

It is quite possible that further useful improvements could be made with some of the simple traditional water lifts. This is an area where study and experimentation, perhaps as part of technical educational programmes or by NGOs working in the field, might yield useful results.

In many cases human powered devices offer the best means to initiate small scale lift irrigation because of their low first cost. However, in the longer term it is to be hoped that small farmers will be assisted to advance towards more productive pumping techniques, which inevitably require some mechanization.

4.2.3    Handpumps

Handpumps, although less productive than footpumps, are the most common form of industrially manufactured manually operated water lift, and for that reason are very widely used. The classic design of piston pump or bucket pump is shown in Figs. 29 and 30. Most of these pumps were developed for use by a family to provide water for themselves and their livestock, rather than for irrigation. The problem when pumps of this kind are used for irrigation is the intensity with which they are used compared with their use for water supply; instead of pumping for a few minutes per day they have to be used for several hours per day, which naturally tends to shorten their useful lives considerably and also to increase the incidence of breakages.

The forces involved in driving a piston pump with a lever have already been discussed in Section 3.5.1. So far as handpumps are concerned, Hofkes [21] indicates the following maximum heads as being generally suitable for comfortable operation of various common sizes of handpump:

The load at any given head can be reduced by shortening the stroke of the pump, but the above recommendation presumably applies with typical strokes in the region of 150 to 300mm.

Hofkes [21] and McJunkin [26] give the nomograph of Fig. 88 as a method for determining handpump discharge. The method for using this is to rule a pencil line between the stroke length that applies (250mm or 10in in the example) and the expected pumping frequency (40 strokes per minute in the example). Then if another pencil line is ruled from where the first line crosses the "pivot line" through the appropriate cylinder  diameter  (76mm  or 3in in the example), the discharge is given on the left (46 litre/min or 12 US gall/min in the example). No allowance is made for "slippage" or leakage of water which will result in the discharge being less than the swept volume, so the result of using this nomograph is the maximum flow that might be expected; it may therefore be more realistic to reduce the result obtained by 10-20%.

Fig. 88 Nomograph for calculating hand pump discharge (after Hofkes [21]and McJunkin[26]).

  There have been many failures of handpumps in the field, especially in water supply projects where no particular individual readily takes responsiblity for the pump. In some countries communal handpumps developed a bad reputation as a result. Mention should be made of the major international UNDP/World Bank Global Handpump Project which seeks among other things to identify good handpumps for development use (mainly for village water supplies); this project has achieved a lot in this field and yielded a number of publications of relevance, notably [27],

Fig. 29 shows some of the key features of a typical handpump. Some of the weak points are the lever and fulcrum mechanism and the pump column itself; these can crack through metal fatigue or due to the use of poor quality castings. The pump in Fig. 29 is good in having a bracing strut to support the pump body, but it is bad in having the pivot bolt for the handle passing through the middle of the most highly stressed part of the pump lever. A better method of pivoting the hand lever is to have the pivot bearing passing through a lug below the lever arm to avoid weakening the arm at that point.

Fig. 89 Rotary drive hand pump (The Gambia)

Fig. 90 Rower pump (part sectioned) (after Klassen [28])

A further problem with hand lever pumps is wear and tear resulting  from "hammering" in the drive train; this can be caused by worn pivots and bearings causing backlash, which will cause impacts if the operator lifts the handle too rapidly so it tries to overtake the piston and pump rod on the down stroke and then suddenly takes up the load again and also by users causing the hand lever to hit the end stops at the end of its travel. Also the need for the operator to constantly raise and lower a heavy lever (plus their arm(s)) wastes energy. Therefore rotary-drive pumps, in which the piston is driven by a crank from a rotating drive wheel are often easier to work (see Figs. 31 and 89). Here a flywheel smooths the fluctuations and thereby makes the pump easier to operate, especially for long periods, because the cyclic loading involved in accelerating discrete cylinder-volumes of water up the rising main will be absorbed by the flywheel's momentum and therefore not be felt by the operator. Rotary drive pumps of this kind have a further advantage as they cannot suffer from damage common with lever pumps from the operator hammering the end stops by moving the lever to far, or by the pump rod momentarily going "slack" due to the operator lifting the hand lever to rapidly.

The main disadvantage with rotary drive pumps is that they are generally relatively heavy and expensive due to  their massive  flywheel  and crank mechanism, plus the supporting column that  is needed.  However, for pumping large quantities of water, as are required for irrigation, the improved ergonomics of the rotary drive is probably more advantageous than it is with water supply duties where no individual is likely to need to pump for more than a few minutes at any one time. 

 Ideally, any piston pump used for irrigation needs to use more than  just the arm muscles. An attempt to produce such piston pump has been made in Bangladesh by the Mennonite Central Committee, with help from Caritas, according to Klassen; [28]. This pump is known as the "Rower Pump" because it is inclined at around 30° to the horizontal and operated with a rowing action (see Fig. 90). It is basically a simple and cheaply constructed pump, using 2" PVC pipe as the pump cylinder. The Rower Pump is claimed to pump 50% more water than a standard UNICEF No.6 lever hand pump used in MOSTI installations, (Manually Operated Shallow Tubewell for Irrigation), and the improvement is proportionately greater with higher lifts. However the main performance advantage of the rower pump is attributed by Klassen not so much to the action which is used (although this is claimed to make it easier to operate and might therefore allow longer periods of use per day) but because of a suction air chamber which smooths the flow into the pump (see Sections 3.5.4 and 3.7.2 for explanations of the use of air-chambers). Without the air chamber the performance falls to a level almost identical with that of the UNICEF pump, while if an air-chamber is fitted on the suction side of the UNICEF pump its performance is also enhanced to a similar level to that of the Rower Pump. Average performance quoted by Klassen for these two pumps, when pumping through heads in the range 5-6.5m (17-21ft), is as follows:

This work is a good example of the kind of useful innovative developments that can be successfully pursued by NGOs and educational institutions in this important field, as suggested earlier. It would be interesting to see the effect of adding air-chambers to the performance and ease of use of other types of handpump in use in the field.

4.2.4     Handpump Maintenance

The problem with industrially manufactured handpumps, as opposed to "self-build" devices like dhones or counterpoise lifts, is that they are dependent on spare parts which cannot easily be locally improvised. They also tend to need a certain amount of preventive maintenance if premature failure of components or impaired performance is to be avoided. However, they probably represent one of the only routes readily available to channel development funds into the widespread deployment of water-lifting equipment, since the alternative of using traditional hand-made devices is usually only practicable where there is an existing tradition make them and where the conditions for their use are right (eg. very low lifts from surface water).

Many water supply programmes using large numbers of handpumps have suffered serious difficulties with pump failures which have generally been attributed to poor maintenance. In some cases the failures have also been due to the use of poor pump designs which lack the capacity to survive intensive use.  The kinds of problems experienced with hand pumps include:

1. poor quality of pump design and manufacture. This has partly resulted from manufacturers trimming the weight (and hence the cost) of components and generally degrading well tried designs in seeking to offer acceptably low bids in the absence of proper specifications to the procurement agencies;
2. iron and steel plain bearings and journals with poor fits and large clearances are provided; these properly should require very frequent lubrication which is impossible to provide, so rapid wear occurs;
3. the great variety of pumps in use leads to difficulties in finding the right spare parts;
4. very limited record keeping and feedback from the field makes it difficult to analyse the reasons for failures and to introduce remedial measures;
5. limited maintenance skills and equipment make it difficult for local people to undertake even basic overhaul operations,  while lack of transport and poor communications make it difficult to summon help from a central source.

Attempts have been made to overcome some of these problems by introducing either a centralized system in which a maintenance team tours around repairing and maintaining a few dozen pumps in a district (this has often proved ineffective and expensive in practice). The other option being advocated is a "two tier" maintenance system, in which a central agency carries out the original installation, and provides a source of spare parts, training, transport, etc., but local people are trained to carry out routine repairs and maintenance.

4.3    ANIMAL POWER

The advantages of animal power over human power are twofold. First; draft animals are five to ten times more powerful than humans, so they can pump more water in a shorter time which tends to make the irrigation operation more efficient and productive. Second; by freeing the operator from having to work the water-lifting device, he can often manage the water distribution system more effectively. In effect, the use of an animal provides the equivalent power of several people, generally at a fraction of the cost.

Some 200 million draft animals are deployed in developing countries [29], and these have an aggregate power capacity of about 75 000 000kW and a capital value in the region of US $20 billion. The majority of these animals are used in southern and south east Asia; 80 million draft animals are in use in India alone. Any mechanization programme to replace these animals would obviously have to be extremely large; but there is more immediate scope for improving the efficiency with which animals are used.

Animal powered irrigation is almost exclusively practised using traditional water lifting techniques pre-dating the industrial era. Although some attempts to produce improved mechanisms for the utilisation of animal power have been made in certain areas during this century, there is little (if any) tendency to introduce animal powered water lifting anywhere where it has not been traditionally practised, although there seems good reason to believe it could usefully replace human labour for irrigation in many parts of the world where it is not already used. Instead, the trend has been either to make the quantum leap to full mechanisation using engines or mains electricity, or to make no attempt to improve on human power. An interesting exception is the "Water Buffalo Project" in Thailand where the use of buffalo powered persian wheels is being encouraged, [24].

The main disadvantage of animal power is that animals need to be fed for 365 days of the year, yet the irrigation season usually only extends for 100 or at most 200 days of the year. In areas where water is close to the surface, and hence irrigation by animal power is feasible, there are usually high human population densities and a shortage of land. Since draft animals consume considerable volumes of fodder, a significant proportion of the available land can be absorbed simply to support the draft animals. Therefore it probably would be difficult to justify the use of animals for irrigation pumping alone, but generally there are other economic applications for them, such as transport, tillage, and post-harvest duties like threshing or milling which allows them to be employed more fully than if they were used exclusively for irrigation. In India and other countries where animal powered water lifting is widely practised, it is normal for the same animals to be used for transport and for tilling the land. They are often fed with agricultural residues, or are allowed to graze on fields left fallow for a season as part of a crop rotation.

Animals also serve other purposes; they are a non-monetary form of collateral which is important in the village economy in many regions, they produce byproducts such as leather, meat and milk and of course in the Indian sub-continent in particular, their dung is widely used as a cooking fuel. So mechanization of irrigation pumps in lieu of draft animals is not necessarily a straightforward alternative.

4.3.1     Power capabilities  of various  species

Fig. 91 gives the approximate water lifting capability of different numbers of oxen, assuming 60% efficiency from animal to water lifted, as might be achieved by one of  the better types  of water  lifting device.

Fig. 91 The number of oxen required to provide a specified quantity of water at different lifts. The curves have been calculated by assuming than an ox can provide 350 watts of power for 5 hours per day, and the efficiency of the water-lifting device is 60%

 The typical power capabilities of various commonly used draft animals  are given in Table 13, based on [1] [30] [24] and other sources.

Draft animals obviously require rest just as humans do, so it is common practice to work them on about three hour shifts, with a rest in between.  10 or 12 hours per day in total may be worked when necessary.

Hood, [31], discussing typical draft horses as used in the USA at the end of the last century advocated the use of efficient mechanisms for coupling horses to water-lifts. He suggested that a team of two horses with an efficient water lifting device can typically lift 0.22 acre.feet per day through a head of 15ft, (270m³/day through 4.6m). He also cited various experiments completed at that time in Madras, India, (using bullocks), as an example for American farmers to emulate. In one of these, a device known as the Stoney Water Lift, was tested and shown to have an animal-to-water mechanical efficiency of about 80%. This device was, in effect, an engineered version of the traditional circular mohte with two balanced buckets, (see Fig. 93). Using a single Nellore bullock in a series of tests, it lifted between 2000 and 2500 Imp.Gall/hr through 22-23ft (9-11.25m³/h through 6.7-7.0m). It was pointed out that although a high instantaneous efficiency was achieved by this two bucket lift, in the tests it was found that the animals were only actually performing useful work for 60% of the duration of the tests, so the utilisation was not as high as may theoretically be feasible; clearly the management of an animal powered water lift has a major influence on the daily productivity.

An important point emphasised by Hood is that all draft animals work best when subjected to a steady load which matches their pulling capability, (although a horse, for example, can throw about one third of its weight into pulling for short periods - i.e. approaching three times the force it can sustain for a long time). Therefore, devices are needed which shield the animal from any cyclic loadings such as is experienced when a pump is driven by a crank, (see explanation of cyclic torque requirements of piston pump in Section 3.5).

4.3.2    Food Requirements

Birch and Rydzewski, [30], show that a cow in Bangladesh can be fed solely with forage and agricultural residues, (although the latter have a value equivalent to US $0.30 per day (1980) which is not a trivial sum) and that the same animal requires the residues from 0.77ha of double-cropped agricultural land. Because of the general shortage of land and residual fodder in Bangladesh, it is unusual to use animals for water lifting in that country.

The same authors carried out a similar calculation in relation to Egypt, where one hectare of land is needed to produce sufficient residue to support one animal, and the daily diet is valued at US $0.50.

4.3.3 Coupling animals to water lifting systems

The original method of using animals to lift water was some device such as the mohte, (Figs. 22 and 92). Here animals walk in a straight line down a slope away from the well or water source while hauling water up in a bag or container. Traditional mohtes used a leather bag to collect the water, but in recent years more durable materials such as rubber truck inner-tubes (or more rarely steel oil barrels) have been used.

Fig. 92 Cross-section view of a Mohte (after Schioler [24])

Fig. 93 Circular Mohte utilizing two buckets with flap-valves in bottom

The mohte is simple to implement; the only mechanical component is a pulley and the only structure is the frame to hold the pulley. However this key mechanical component is more complicated than might be expected due to the load on it, and demands considerable craftsmanship [24]. The instantaneous efficiency of the mohte is high while the animal(s) are pulling; a disadvantage is the need to reverse the animals back up the ramp to lower the bucket. Sometimes two teams of animals are used, so that one is led back to the top while one team descends; in order to do this, a man is needed at both ends of the ramp to harness and un-harness the animals at the end of each cycle. Using two sets of animals and, generally, two men and a boy, practically doubles the output (Molenaar, [1]), but even so inevitably no action occurs during the harnessing and un-harnessing process.

The downhill slope helps use the weight of the animal partially to balance the weight of the full water container; it also applies a reasonably constant load to the animal for most of the journey down the ramp, from when the container leaves the surface of the water to when it starts to be tipped.

 An improved version of the mohte, sometimes used in parts of India and (a few) in Sri Lanka, is the "circular mohte"; Fig. 93. This involves attaching the animals to a sweep so that they can walk in a circle thereby allowing the animals to work continuously with less supervision. Because their weight can no longer partially balance the load, as they have to walk on level ground, two buckets are used so that the empty one descends while the full one comes up; this at least balances the weights of the buckets and means that the water being lifted is the only large out-of-balance force to be handled. The main.problem with this device is that the load on the sweep is cyclic, as the pull on the sweep by the chain will not be felt by the animals when the chain acts parallel to the sweep and it will reach a maximum when the chain is at right angles to the sweep; therefore the animals will have a tiring sinusoidal load to cope with. Also, the various pulleys and supporting posts need to be robust and well anchored, as the forces are quite large.

 The Persian wheel, (Figs. 23, 94 and 95) is a great improvement on the mohte, as its chain of buckets imposes an almost constant load on the drive shaft to the wheel. Persian wheels are usually driven by some form of right angle drive, such as in Figs. 94 and 95. The first is the most common, where the drive shaft from the secondary gear is buried and the animals walk over it; this has the advantage of keeping the Persian wheel as low as possible to minimise the head through which water is lifted. The second example is a traditional wooden Persian wheel mechanism where the animal passes under the horizontal shaft. The sweep of a Persian wheel carries an almost constant load and therefore the animal can establish a steady comfortable pace and needs little supervision.

In Egypt, the sakia (see Section 3.4.3 and Fig. 26) is commonly used for low head applications instead of the Persian wheel, and it is driven in a similar manner via a sweep. It has the advantage of applying a constant load to the animal, but is more efficient at very low heads.

The next development was for the sweep drive gear to be "industrialised" and manufactured in large  numbers  from iron or steel to include an engineered set of gears. Fig. 96 shows a mule harnessed to such a device in order to drive a chain and washer (or paternoster) pump which are widely used in China, where millions of liberation pumps, many of which were animal powered, were produced as an intermediate stage between human pumping and full mechanisation. The liberation pump includes a mechanism driven by a sweep that is elegantly simple and made from steel castings; (see Figs. 53 and 90). As detailed in Section 3.6.7, the liberation pump is capable of achieving an efficiency of approaching 70% in the animal powered form illustrated, and it is compact enough to fit into quite a narrow well.

Fig. 94 A bullock-driven Persian wheel of the conventional chain and bucket type which is widely used in many parts of the world, particularly the Near East and Southern Asia

Fig. 95 Camel-driven Persian wheel showing over-head drive mechanism (ref. Schioler [24])

Fig. 96 Animal-powered Chinese Liberation Pump A hand-operated version is shown in Fig. 53 and a motorized one in Fig. 84 ,

Another not unusual concession to modernisation is the use of an old motor vehicle back-axle embedded in a concrete pillar as a means of obtaining a right angle drive from an animal sweep; an example of this is indicated in Fig. 97, where a donkey is shown linked to an Archimedian screw, (see also Section 3.6.3 and Fig. 48). Although in the example illustrated the matching problem is not always solved satisfactorily, according to Schioler, [24], Again this is a compact and potentially efficient mechanism, as the Archimedean screw applies a completely steady load to the animal. It is clear that the same mechanism improvised from a car back axle could equally easily be coupled to a sakia, to a Persian wheel or to a chain and washer pump.

A new development in this field has been the appearance of a prototype animal powered, but industrially manufactured double-acting diaphragm pump from Denmark. This has the advantage of having the sweep direct coupled to the pump, which acts as a suction pump. Therefore it is only necessary to bury the pipes carrying the water rather than a drive shaft and this device can be located up to 80m away from the water source, which may in some cases be an advantage. This "Bunger" sweep pump is claimed to lift 100m³/day using two animals.

Fig. 97 The back axle from a car used as an animal-driven power transmission for an Archimedean screw pump (see also Fig. 48)

Although a sweep driven device avoids the problem of reversing animals as with a mohte, it suffers from the disadvantage that by forcing an animal to walk in a circle, even though the load may be steady, the tractive effort or pull is reduced to 80% of that which is feasible when the same animal walks in a straight line; Hood; [31]. A mechanism which can apply a steady load to the animal, but in a straight line, would be better. The simplest device of this kind is the tread-wheel; in some respects the principle is analagous to a mohte, with the animal "walking on the spot" inside a large wheel, rather than up and down a ramp. This principle was taken further at the end of the nineteenth century in Europe and the USA through the use of "Paddle wheel" animal engines, in which a horse would be harnessed on an inclined endless belt which it would drive with its feet. The disadvantage of these animal carrying devices is that a mechanism is needed which not only has to transmit the maximum draw-bar pull of the animal, but additionally has to carry the full weight of the beast. Therefore a massive and robust construction is necessary which inevitably is expensive, and this probably is an example of where the search for maximum efficiency produces diminishing returns and therefore is counter-productive.