3.7    RECIPROCATING INERTIA (JOGGLE) PUMPS

This range of pumps depend on accelerating a mass of water and then releasing it; in other words, on "throwing" water. They are sometimes known as "inertia" pumps.

As with the other families of pumps so far reviewed, there are both reciprocating inertia pumps, described below, (which are only rarely used) and much more common rotary types which include the centrifugal pump, described in Section 3.8.

3.7.1    Flap Valve Pump

This is an extremely simple type of pump which can readily be improvized; (see Fig. 54). Versions have been made from materials such as bamboo and the dimensions are not critical, so that little precision is needed in building it.

The entire pump and riser pipe are joggled up and down by a hand lever, so that on the up-stroke the flap valve is sucked closed and a column of water is drawn up the pipe, so that when the direction of motion is suddenly reversed the column of water travels with sufficient momentum to push open the flap valve and discharge from the outlet. Clearly a pump of this kind depends on atmospheric pressure to raise the water, so it is limited to pumping lifts of no more than 5-6m.

3.7.2    Resonant Joggle Pump

Fig. 55 shows an improved version of the flap-valve pump. Here there is an air space at the top of the pump which interacts with the column of water by acting as a spring, to absorb energy and then use it to expel water for a greater part of the stroke than is possible with a simple flap-valve pump. This uses exactly the same principle as for an air chamber (see Section 3.5.4).

Fig. 54 Flap valve pump

Fig. 55 Joggle pump

The joggle pump depends on being worked at the correct speed to make it resonate. An example of a resonant device is a weight hanging from a spring, which will bounce up and down with a natural frequency determined by the stiffness of the spring and the magnitude of the weight. The heavier the weight in relation to the spring stiffness, the slower the natural frequency and vice-versa. If the spring is tweaked regularly, with a frequency close to its natural frequency, then a small regular pull applied once per bounce can produce a large movement quite easily, which is an example of resonance. In exactly the same way, each stroke of a resonant joggle pump makes a column of water of a certain mass bounce on the cushion of air at the top of the column. Depending on the size of the air chamber and the mass of the water, this combination will tend to bounce at a certain resonant frequency. Once it has been started, a pump of this kind needs just a regular "tweak" of the handle at the right frequency to keep the water bouncing. This effect not only improves the overall efficiency but makes it relatively effortless to use. Dunn [20] reports performance figures of 60 to 100 litres/minute lifted through 1.5 to 6m at a frequency of 80 strokes per minute.

It is worth noting that the performance of some reciprocating piston pumps fitted with airchambers (as in Fig. 36 C) can be similarly enhanced if the speed of the pump is adjusted to match the resonant frequency of the water in the pipeline and the "stiffness" of the trapped air in the air-chamber. This is usually only feasible with short pipelines at fairly low heads, as otherwise the natural frequency in most practical cases is far too low to match any reasonable pump speed. If resonance is achieved in such situations the pump will often achieve volumetric efficiencies in the region of 150 to 200%; i.e. approaching twice the swept volume of the pump can be delivered. This is because the water continues to travel by inertial effects even when the pump piston is moving against the direction of flow, (the valves of course must remain open). As a result, water gets delivered for part of the down stroke as well as on the up stroke. Well-engineered reciprocating systems taking advantage of resonance can achieve high speeds and high efficiencies. Conversely, care may be needed in some situations (such as pumps where there is a reversal of the direction of flow), to avoid resonance effects, as although they can improve the output, they can also impose excessive loads on the pump or on its drive mechanism.

3.8    ROTODYNAMIC PUMPS

3.8.1    Rotodynamic Pumps: Basic Principles

The whole family of so-called rotodynamic pumps depends on propelling water using a spinning impeller or rotor. Two possible mechanisms are used either alone or in combination, so that water is continuously expelled from the impeller by being:

1. deflected by the impeller blades (in propeller type pumps);
2. whirled into a circular path so centrifugal force then carries the water away, in the same way a weight on a string when whirled around and released will fly away.

The earliest practical rotodynamic pumps were developed in the 18th and early 19th century, (Fig. 56). Type A in the figure simply throws water outwards and upwards. Type B is actually a suction centrifugal pump and needs priming in order to initiate pumping; a foot valve is provided to prevent the loss of the priming water when the pump stops. A circular casing is provided to collect the output from the impeller at the delivery level. A pump of this kind is extremely inefficient as the water leaves the impeller with a high velocity which is simply dissipated as lost energy. Pump C, the Massachusetts Pump of 1818, had the collector built around a horizontal shaft so that the velocity of the water could be directed up the discharge pipe and carry it to some height; in some respects this is the fore-runner of the modern centrifugal pump which today is the most commonly used mechanically driven type of pump.

3.8.2    Volute, Turbine and Regenerative Centrifugal Pumps

The early pumps just described differed from modern pumps in one important respect; the water left the pump impeller at high speed and was only effectively slowed down by friction, which gives them poor efficiency and poor performance. The application of an important principle, shown in Fig. 57, led to the evolution of efficient rotodynamic pumps; namely that with flowing fluids, velocity can be converted into pressure and vice-versa. The mechanism is to change the cross section of the passage through which water (or any liquid) is flowinq. Because water is virtually incompressible, if a given flow is forced to travel through a smaller cross section of passage, it can only do so by flowing faster. However pressure is needed to create the force needed to accelerate the mass of water. Conversely, if a flow expands into a larger cross section, it slows down to avoid creating a vacuum and the deceleration of the fluid imposes a force and hence an increase in pressure on the slower moving fluid. It car. be shown (if frictional effects are ignored) that if water flows through a duct of varying cross sectional area, then the head of water (or pressure difference) to cause the change? in velocity from v, to ^v_out, will be H, where:

where g is the acceleration due to gravity.

Fig. 56 Early types of centrifugal pumps

Fig. 57 The relationship between pressure and velocity through both a jet and a diffuser

The diagram in Fig. 57 shows how the pressure decreases in a jet as the velocity increases while the reverse occurs in a diffuser which slows water down and increases the pressure. Qualitatively this effect, is obvious to most people. From experience, it is well known that pressure is needed to produce a jet of water; the opposite effect, that smoothly slowing down a jet increases the pressure is less obvious.

When this was understood, it became evident that the way to improve a centrifugal pun; is to throw the water out of an impeller at high speed (in order to add the maximum energy to the water) and then to pass the water smoothly into a much larger cross section by way of a diffuser in which the cross section changes slowly. In this way, some of the velocity is converted into pressure. A smooth and gradual change of cross-section is essential, any sudden change would create a great deal of turbulence which would dissipate the energy of the water instead of increasing the pressure. There are two main methods of doing this, illustrated in Fig. 58 by diagrams A and B, and a more unusual method shown in C.

Diagram A shows the most common, which is the "volute centrifugal" pump, generally known more simply just as a "centrifugal" pump. Here a spiral casing with an outer snail-shell-shaped channel of gradually increasing cross section draws the output from the impeller tangentially, and smoothly slows it down. This allows the water to leave tangentially through the discharge pipe at. reduced velocity, and increased pressure.

Diagram B shows the other main alternative, which is the so-called "turbine centrifugal" or "turbine pump", where a set of smoothly expanding diffuser channels, (six in the example illustrated) serve to slow the water down and raise its pressure in the same way. In the type Of turbine pump illustrated, the diffuser channels also deflect the water into a less tangential and more radial path to allow it to flow smoothly into the annular constant cross-section channel surrounding the diffuser ring, from whore it discharges at the top.

Diagram. C shows the third, lesser known type of centrifugal pump which is usually called a "regenerative pump", but is also sometimes called a "side-chamber pump" or ever, (wrongly) a "turbine pump". Here an impeller with many radial blades turns in a rectangular sectioned annulus; the blades accelerate the water by creating two strong rotating vortexes which partially interact with the impeller around the rim of the pump for about three-quarters of a revolution; energy is steadily added to the two vortexes each time water passes through the impeller; for those familiar with motor vehicle automatic transmissions the principle is similar to that of the fluid flywheel. When the water leaves the annulus it passes through a diffuser which converts its velocity back into pressure. Regenerative pumps are mentioned mainly for completeness; because they have very close internal clearances they are vulnerable to any suspended grit or dirt and are therefore only normally used with clean water (or other fluids) in situations where their unique characteristics are advantageous. They are generally inappropriate for irrigation duties. Their main advantage is a better capability of delivering water to a higher head than other types of single-stage centrifugal pump.

Fig. 58 Centrifugal pump types

3.8.3    Rotodynamic Pump Characteristics and Impeller Types

It is not intended to deal with this complex topic in depth, but it is worth running through some of the main aspects relating to pump design to appreciate why pumps are generally quite sensitive to their operating conditions.

All rotodynamic pumps have a characteristic of the kind illustrated in Fig. 16, which gives them a limited range of speeds, flows and heads in which good efficiency can be achieved. Although most pumps will operate over a wider range, if you move far enough from their peak efficiency with any of these parameters, then both the efficiency and output will eventually fall to zero. For example, Fig. 16 shows that if you drive the pump in question with a motor having a maximum speed of say 2000 rpm, there is a maximum flow which can be achieved even at zero head, and similarly there is a head beyond which no flow will occur. The design point is usually at the centre of the area of maximum efficiency.

Since any single rotodynamic pump is quite limited in its operating conditions, manufacturers produce a range of pumps, usually incorporating many common components, to cover a wider range of heads and flows. Because of the limited range of heads and flows any given impeller can handle, a range of sub-sets of different types of impellers has evolved, and it will be shown later there are then variations within each sub-set which can fine tune a pump for different duty requirements. The main sub-sets are shown in Fig. 59, which shows a half-section through the impellers concerned to give an idea of their appearance.

Fig. 59 Typical rotodynamic pump characteristics

Fig. 60 Axial flow (or propeller) pump

It can be seen that pump impellers impose radial, or axial flow on the water, or some combination of both. Where high flows at low heads are required (which is common with irrigation pumps), the most efficient impeller is an axial flow one (this is similar to a propeller in a pipe) - see Fig. 60. Like a propeller, this depends on lift generated by a moving streamlined blade; since in this case the propeller is fixed in a casing, the reaction moves the water. Conversely, for high heads and low flows a centrifugal (radial flow) impeller is needed with a large ratio between its inlet diameter and its outlet diameter, which produces a large radial flow component, as in the left-most type in Fig. 59. In between these two extremes are mixed flow pumps (see also Figs. 61 and 62) and centrifugal pumps with smaller ratios of discharge to inlet diameter for their impellers. The mixed flow pump has internal blades in the impeller which partially propel the water, as with an axial flow impeller, but the discharge from the impeller is at a greater diameter than the inlet so that some radial flow is involved which adds velocity to the water from centrifugal forces that are generated.

Fig. 59 also shows the efficiency versus the "Specific Speed" of the various impeller sub-sets. Specific Speed is a dimensionless ratio which is useful for characterising pump impellers (as well as hydro-turbine rotors or runners). Text books on pump/turbine hydrodynamics cover this topic in greater depth. The Specific Speed is defined as the speed in revolutions per minute at which an impeller would run if reduced in size to deliver 1 litre/sec to a head of 1m and provides a means for comparing and selecting pump impellers and it can be calculated as follows:

where n is speed in rpm, Q is the pump discharge in litre/sec and H is the head in metres.

Fig. 61 Surface mounted mixed flow pump

Fig. 62 Submerged mixed flow pump

Fig. 59 indicates the Specific Speeds which best suit the different impeller sub-sets; e.g. an axial flow impeller is best at flow rates of 500-1 000 litre/sec and has a Specific Speed of 5 000-10 000, at heads of about 5m. Specific Speed can be converted back to actual rpm (n) at any given head (H) and flow (Q) as follows:

where n is in rpm, N is the Specific Speed from Fig. 59, H is the head in metres, and Q is the flow in litres/sec.

The choice of impeller is not only a function of head and flow but of pump size too; smaller low powered pumps of any of these configurations tend to be somewhat less efficient and they also operate best at lower heads than geometrically similar larger versions.

Fig. 59 also indicates the effect on power requirements and efficiency (marked "kW" and "EFF" respectively) of varying the key parameter of head "H", away from the design point. In the case of a centrifugal pump the small diagram shows that increasing the head reduces the power demand, while in the case of an axial-flow pump, increasing the head increases the power demand. Paradoxically, reducing the head from the design head on a centrifugal pump increases the power demand; the reason for this is that decreasing the head by, say, 10% can increase the flow by 25% - the efficiency may also go down by 10%, and since the power requirement is head times flow divided by efficiency, the new power demand will change from:

the ratio of these is 1:1.25, so the power demand will be increased by 25% in this case. Therefore, varying the conditions under which a pump operates away from the design point can have an unexpected and sometimes drastic effect. The use of pumps off their design point is a common cause of gross inefficiency and wasted fuel.

3.8.4    Axial-Flow (Propeller) Pumps

As already explained, an axial-flow (or "propeller") pump propels water by the reaction to lift forces produced by rotating its blades. This action both pushes the water past the rotor or impeller and also imparts a spin to the water which if left uncorrected would represent wasted energy, since it will increase the friction and turbulence without helping the flow of water down the pipe. Axial flow pumps therefore usually have fixed guide vanes, which are angled so as to straighten the flow and convert the spin component of velocity into extra pressure, in much the same way as with a diffuser in a centrifugal pump. Fig. 60 shows a typical axial flow pump of this kind, in which the guide vanes, just above the impeller, also serve a second structural purpose of housing a large plain bearing, which positions the shaft centrally. This bearing is usually water lubricated and has features in common with the stern gear of an inboard-engined motor boat.

Axial flow pumps are generally manufactured to handle 'flows in the range 150 to 1 500m³/h for vertically mounted applications, usually with heads in the ranqe 1.5-3.On. By adding additional stages (i.e. two or more impellers on the same shaft) extra lift up to 10m or so can be engineered.

Because pumps of this kind are designed for very large flows at low heads, it is normal to form the "pipes" in concrete as illustrated, to avoid the high cost of large diameter steel pipes. Most axial flow pumps are large scale devices, which involve significant civil works in their installation, and which would generally only be applicable on the largest land-holdings addressed by this publication. They are generally mainly used in canal irrigation schemes where large volumes of water must be lifted 2-3m, typically from a main canal to a feeder canal.

Small scale propeller pumps are quite successfully improvised but not usually manufactured; ordinary boat propellers mounted on a long shaft have been used for flooding rice paddies in parts of southeast. Asia. The International Rice Research Institute (TRRI) has developed this concept into a properly engineered, portable high volume pumping system, (see Fig. 63); it is designed to be manufactured in small machine shops and is claimed to deliver up to 180m³/h at heads in the range l-4m. This pump requires a 5hp (3kW) engine or electric motor capable of driving its shaft at 3 000rpm; its; length is 3.7m, the discharge tube is 150mm in diameter and the overall mass without the prime mover fitted is 45kg.

Fig. 63 Portable axial flow pump (IRRI)

3.8.5    Mixed-Flow Pumps

The mixed-flow pump, as its name suggests, involves something of both axial and centrifugal pumps and in the irrigation context can often represent a useful compromise to avoid the limited lift of an axial flow pump, but still achieve higher efficiency and larger flow rates than a centrifugal volute pump. Also, axial flow pumps generally cannot sustain any suction lift, but mixed-flow pumps can, although of course they are not self-priming.

Fig. 61 shows a surface mounted, suction mixed-flow pump and its installation. Here the swirl imparted by the rotation of the impeller is recovered by delivering the water into a snail-shell volute or diffuser, identical in principle to that of a centrifugal volute pump.

An alternative arrangement more akin to an axial flow pump is shown in Fig. 62. Here what is often called a "bowl" casing is used, so that the flow spreads radially through the impeller, and then converges axially through fixed guide vanes which remove the swirl and thereby, exactly as with axial flow pumps, add to the efficiency. Pumps of this kind are installed submerged, which avoids the priming problems that can afflict large surface suction rotodynamic pumps such as in Fig. 61. The "bowl" mixed-flow pump is sometime called a "turbine" pump, and it is in fact analagous to the centrifugal turbine pump described earlier; the passage through the rotor reduces in cross-section and serves to accelerate the water and impart energy to it, while the fixed guide vanes are designed as a diffuser to convert speed into pressure and thereby increase both the pumping head and the efficiency. A number of bowl pumps can be stacked on the same shaft to make a multi-stage turbine pump, and these are quite commonly used as borehole pumps due to their long narrow configuration. Mixed-flow bowl pumps typically operate with flows from 200-12 000m³/h over heads from 2-10m. Multiple stage versions are often used at heads of up to about 40m.

3.8.6    Centrifugal Pumps

i. Horizontal shaft centrifugal pump construction

These are by far the most common generic type of electric or engine powered pump for small to medium sized irrigation applications. Fig. 64 shows a typical mass-produced volute-centrifugal pump in cross section. In this type of pump the casing and frame are usually cast iron or cast steel, while the impeller may be bronze or steel. Critical parts of the pump are the edges of the entry and exit to the impeller as a major source of loss is back-leakage from the exit of the impeller around the front of it to the entry. To prevent this, good quality pumps, including the one in the diagram, have a closely fitting wear ring fitted into the casing around the front rim of the impeller; this is subject to some wear by grit or particulate matter in the water and can be replaced when the clearance becomes large enough to cause significant loss of performance. However, many farmers probably do not recognize wear of this component as being serious and simply compensate by either driving the pump faster or for longer each day, both of which waste fuel or electricity. Another wearing part is a stuffing box packing where the drive shaft emerges from the back of the impeller casing. This needs to be periodically tightened to minimize leakage, although excessive tightening increases wear of the packing. The packing is usually graphited asbestos, although graphited PTFE is more effective if available. The back of the pump consists of a bearing pedestal and housing enclosing two deep-groove ball-bearings. This particular pump is oil lubricated, it has a filler, dip-stick and drain plug. Routine maintenance involves occasional changes of oil, plus more frequent checks on the oil level. Failure to do this leads to bearing failure, which if neglected for any time allows the shaft to whirl and damage the impeller edges.

Fig. 64 Typical surface mounted pedestal centrifugal pump

ii . Centrifugal pump installations

Figs. 65 and 66 show two alternative typical low lift centrifugal pump installations; the simplest is the suction installation of Fig. 65. As mentioned earlier in Section 2.1.5, centrifugal pumps are limited to a maximum in practice of about 4-5m suction lift at sea level (reducing to around 2m suction lift at an altitude of 2 000m, and further reduced if a significant length of suction pipe is involved; otherwise problems are almost certain to be experienced in priming the pump, retaining its prime, etc. A foot valve is a vital part of any such installation as otherwise the moment the pump stops or slows down, all the water in the pipeline will run back through the pump making it impossible to restart the pump unless the pipeline is first refilled. Also, if water flows back through the pump, it car: run backwards and possibly damage the electrical system.

If the delivery pipeline is long, it is also important to have another check valve (non-return valve) at the pump discharge to the pipeline. The reason for this is that if for any reason the pump suddenly stops, the flow will continue until the pressure drops enough to cause cavitation in the line; when the upward momentum of the water is exhausted, the flow reverses and the cavitation bubbles implode creating severe water hammer. Further severe water hammer occurs when the flow reverses causing the footvalve to slam shut. The impact of such events has been known to burst a centrifugal pump's casing. The discharge check valve therefore protects the pump from any such back surge down the pipeline.

Fig. 65 Surface mounted centrifugal pump installation

Fig. 66 Below-surface (sump) centrifugal pump installation

Fig. 67 Various types of centrifugal pump impellers

Fig. 68 Effect of direction of curvature of vanes of centrifugal pump impellers

In many cases there is no surface mounting position low enough to permit suction pumping. In such cases centrifugal pumps are often placed in a sump or pit where the suction head will be small, or even as in Fig. 66 where the pump is located below the water level. In the situation illustrated a long shaft is used to drive the pump from a surface mounted electric motor; (to keep the motor and electrical equipment above any possible flood level).

iii. Centrifugal pump impeller variations

The component that more than anything else dictates a centrifugal pump's characteristics is its impeller. Fig. 67 shows some typical forms of impeller construction. Although the shape of an impeller is important, the ratio of impeller exit area to impeller eye area is also critical (i.e. the change of cross section for the flow through the impeller), and so is the ratio of the exit diameter to the inlet diameter. A and B in the figure are both open impellers, while C and D are shrouded impellers. Open impellers are less efficient than shrouded ones, (because there is more scope for back leakage and there is also more friction and turbulence caused by the motion of the open blades close to the fixed casing), but open impellers are less prone to clogging by mud or weeds. But shrouded impellers are considerably more robust and less inclined to be damaged by stones or other foreign bodies passing through. Arguably, open impellers are less expensive to manufacture, so they tend to be used on cheaper and less efficient pumps; shrouded impellers are generally superior where efficiency and good performance are important.

Also in Fig. 67, A and C are impellers for a single-suction pump, while B and D are for a double-suction pump in which water is drawn in symmetrically from both sides of the impeller. The main advantage of a double-suction arrangement is that there is little or no end thrust on the pump shaft, but double suction pumps are more complicated and expensive and are uncommon in small and medium pump sizes.

The shape of the impeller blades is also of importance. Some factors tend to flatten the HQ curve for a given speed of rotation, while others Steepen it. Fig. 68 shows the effect of backward raked, radial and forward raked blade tips; the flattest curve is obtained with the first type, while the last type actually produces a maximum head at the design point. Generally the flatter the HQ curve, the higher the efficiency, but the faster the impeller has to be driven to achieve a given head. Therefore impellers producing the most humped characteristics tend to be used when a high head is needed for a given speed, but at some cost in reduced efficiency.

iv. Series and parallel operation of centrifugal pumps

Where a higher head is needed than can be achieved with a single pump, two can be connected in series as in Fig. 69 A, and similarly, if a greater output is needed, two centrifugal pumps may be connected in parallel as in Fig. 69 B. The effects of these arrangements on the pump characteristics are illustrated in Fig. 69 C, which shows the changes in head, discharge and efficiency that occur as a percentage of those for a single pump operating at its design point. It is clear that series connection of pumps has no effect on efficiency or discharge but doubles the effective head. Parallel operation does not however normally double the discharge compared to a single pump, because the extra flow usually causes a slight increase in total head (due to pipe friction), which will move the operating point enough to prevent obtaining double the flow of a single pump.

Fig. 69 Combining centrifugal pumps in series or parallel

3.8.7    Multi-stage and Borehole Rotodynamic Pumps

Where high heads are needed, the primary means to achieve this with a single impeller centrifugal pump are either to drive the impeller faster or to increase its diameter. In the end there are practical limits to what can be don in this way, so that either single impeller pumps can be connected in series, or a more practical solution is to use a multiple impeller pump in which the out put from it from one impeller feeds directly, through suitable passages in the casing, to the next, mounted on the same shaft. Fig. 70 shows a 5 stage borehole pump (where limitations on the impeller diameter are caused by the borehole, making multi-staging an essential means to obtain adequate heads). Fig. 44 includes a three stage centrifugal pump, coupled to a turbine as a prime-mover, as another example of multi-staging.

Surface-mounted multi-stage pumps are probably only likely to be of relevance to irrigation in mountainous areas since there are few situations elsewhere where surface water needs to be pumped through a high head. More important from, the irrigation point of view is the vertical shaft multi-stage submersible borehole pump which has an integral submerged electric motor directly coupled to the pump below the pump as in the example of Fig. 70, However, it is possible to get bare-shaft multi-stage borehole pumps in which the pump is driven from the surface via a long drive shaft supported by spider bearings at regular intervals down the rising main; see Fig. 134 (b) or with the motor arranged as for the centrifugal pump in Fig. 66, but. with a vertically mounted multi-stage pump in either a sump or well.

In recent years numerous, reliable, submersible electric pumps have evolved; Fig. 70. Section 4.6 discusses in more detail the electrical implications and design features of this kind of motor. Extra pump stages can be fitted quite easily to produce a range of pumps to cover a wide spectrum of operating conditions. The pump in Fig. 70 is a 5 stage mixed-flow type, and the same figure also shows how, simply by adding extra stages (with increasingly powerful motors) a whole family of pumps can be created capable in the example illustrated of lifting water from around 40m with the smallest unit to around 245m with the most powerful; the efficiency and flow will be similar for all of these options. Only the head and the power rating will vary in proportion to the; number of stages fitted.

Finally, Fig. 71 and Fig. 134 (a) show borehole installations with submersible electric pumps. The pump in Fig. 71 has level sensing electrodes clipped to the rising main, which can automatically switch it off if the level falls too low.

Fig. 70 Multi-stage submersible electric borehole pumps

Fig. 71 Schematic of complete electric submersible borehole pumping installation

3.8.8    Self-priming Rotodynamic Pumps

Rotodynamic pumps, of any kind, will only start to pump if their impellers are flooded with water prior to start-up. Obviously the one certain way to avoid any problem is to submerge the pump in the water source, but this is not always practical or convenient. This applies especially to portable pump sets, which are often important for irrigation, but which obviously need to be drained and re-primed every time they are moved to a new site.

If sufficient water is present in the pump casing, then even if the suction pipe is empty, suction will be created and water can be lifted. A variety of methods are used to fill rotodynamic pumps when they are mounted above the water level. It is, however, most important to note that if the suction line is empty but the delivery line is full, it may be necessary to drain the delivery line in order to remove the back pressure on the pump, to enable it to be primed. Otherwise it will be difficult if not impossible to flush out the air in the system. One way to achieve this is to fit a branch with a hand valve on it at the discharge, which can allow the pump to be "bled" by providing an easy exit for the air in the system.

The most basic method of priming is to rely on the footvalve to keep water in the system. The system has to be filled initially by pouring water into the pipes from a bucket; after that it is hoped that the footvalve will keep water in the system even after the pump is not used for some time. In many cases this is a vain hope, as footvalves quite often leak, especially if mud or grit is present in the water and settles between the valve and its seat when it attempts to close. Apart from the nuisance value when a pump loses its prime, many pumps suffer serious damage if run for any length of time while dry, as the internal seals and rubbing faces depend on water lubrication and will wear out quickly when run dry. Also, a pump running dry will tend to overheat; this will melt the grease in the bearings and cause it to leak out, and can also destroy seals, plastic components or other items with low temperature tolerance.

The two most common methods for priming surface-mounted, engine driven suction centrifugal pumps are either by using a small hand pump on the delivery line as illustrated in Fig. 72, (this shows a diaphragm priming pump which has particularly good suction capabilities) or an "exhaust ejector" may be used; here suction is developed by a high velocity jet of exhaust from the engine, using similar principles to those illustrated in Fig. 57 and described in more detail in Section 3.8.9 which follows.

Several alternatie methods of priming surface suction pumps may be commonly improvized. For example, a large container of water may be mounted above the pump level so water can be transferred between the pump and the tank via a branch from the delivery line with a valve in it. Then when the pump has to be restarted after the pipe-line has drained, the valve can be opened to drain the tank into the pump and suction line. Even the worst footvalves leak slowly enough to enable the system to be started, after which the tank can be refilled by the pump so as to be ready for the next start.

Alternatively, a large container can be included in the suction line, mounted above the level of the pump, which will always trap enough water in it to allow the pump to pull enough of a vacuum to refill the complete suction line. Care is needed in designing an installation of this kind, to avoid introducing air-locks in the suction line.

Fig. 72 Direct-coupled air-cooled diesel engine and pump installation with hand-operated diaphragm pump for priming

Yet another simple method to use, but only if the delivery line is long enough to carry a sufficient supply of water, is to fit a hand-valve immediately after the pump discharge (instead of a non-return valve) so that when the pump is turned off, the valve can be manually closed. Then the opening of this valve will refill the pump from the delivery line to ensure it is flooded on restarting.

Sometimes the most reliable arrangement is to use a special "self-priming" centrifugal pump (Fig. 73). Here, the pump has an enlarged upper casing with a baffle in it. When the pump and suction line are empty, the pump casing has to be filled with water from a bucket through the filler plug visible on top. Then when the pump is started, the water in the casing is thrown up towards the discharge and an eye is formed at the hub of the impeller which is at low pressure; until water is drawn up the suction pipe the water discharged from the top of the pump tends to fall back around the baffle and some of the entrained air carries on up the empty discharge pipe. The air which is discharged is replaced by water drawn up the suction pipe, until eventually the suction pipe fills completely and the air bubble in the eye of the impeller is blown out of the discharge pipe. Once all the air has been expelled, water ceases to circulate within the pump and both channels act as discharge channels. A check valve is fitted to the inlet of the pump so that when the pump is stopped it remains full of water. Then even if the foot valve on the suction line leaks and the suction line empties, the water trapped in the casing of the pump will allow the same self-priming function as described earlier to suck water up the suction line. Hence, pumps of this kind only need to be manually filled with water when first starting up after the entire system has been drained.

Fig. 73 Self-priming centrifugal pump

3.8.9    Self-Priming Jet Pumps

An alternative type of self-priming centrifugal pump uses the fact that if water is speeded up through a jet, it causes a drop in pressure (see Section 3.8.2). Here the pump is fitted into a secondary casing which contains water at. discharge pressure, (see Fig. 74). A proportion of the water from this chamber is bled back to a nozzle fitted into the suction end of the pump casing and directed into the eye of the impeller. Once the pump has been used once (having been manually primed initially) it remains full of water so that on start up the pump circulates water from the discharge through the jet and back into the suction side. As before, air is sucked through and bubbles out of the discharge, while (until the pump primes) the water falls back and recirculates. The jet causes low pressure in the suction line and entrains air which goes through the impeller and is discharged, hence water is gradually drawn up the suction line. As soon as all the air is expelled from the system, most of the discharge goes up the discharge line, but a proportion is fed back to the nozzle and increases the suction considerably compared with the effect of a centrifugal impeller on its own. Therefore, this kind of pump not only pulls a higher suction lift than normal, but the pump can reliably run on "snore" (i.e. sucking a mixture of air and water without losing its prime). This makes it useful in situations where shallow water is being suction pumped and it is difficult to obtain sufficient submergence of the footvalve, or where a water source may occasionally be pumped dry.

This jet pump principle can also be applied to boreholes as indicated in Fig. 75. An arrangement like this allows a surface-mounted pump and motor to "suck" water from depths of around 10-20m; the diffuser after the jet serves to raise the pressure in the rising main and prevent cavitation. Although the jet circuit commonly needs 1.5-2 times the flow being delivered, and is consequently a source of significant power loss, pumps like this are sometimes useful for lifting sandy or muddy water as they are not so easily clogged as a submerged pump. In such cases a settling tank is provided on the surface between the pump suction and the jet pump discharge to allow the pump to draw clearer water.

Fig. 74 Schematic of a surface-suction jet pump

The disadvantages of jet pumps are, first, greater complexity and therefore cost, and second, reduced efficiency since power is used in pumping water through the jet, (although some of this power is recovered by the pumping effect of the jet). Obviously it is better to use a conventional Centrifugal pump in a situation with little or no suction lift, but where Suction pumping is essential, then a self-priming pump of this kind can offer a successful solution.

Fig. 75 Borehole jet pump installation

3.9    AIR LIFT PUMPS

The primary virtue of air lift pumps is that they are extremely simple. A rising main, which is submerged in a well so that more of it is below the water level than above it, has compressed air blown into it at its lowest point (see Fig. 76). The compressed air produces a froth of air and water, which has a lower density than water and consequently rises to the surface. The compressed air is usually produced by an engine driven air compressor, but windmill powered air compressors are also used. The principle of it is that an air/water froth, having as little as half the density of water, will rise to a height above the water level in the well approximately equal to the immersed depth of the rising main. The greater the ratio of the submergance of the rising main to the static head, the more froth will be discharged for a given supply of air and hence the more efficient an air lift pump will be. Therefore, when used in a borehole, the borehole needs to be drilled to a depth more than twice the depth of the static water level to allow adequate submergence.

Fig. 76 Air lift pump (schematic)

The main advantage of the air lift pump is that there are no mechanical below-ground components, so it is essentially simple and reliable and can easily handle sandy or gritty water. The disadvantages are rather severe; first, it is inefficient as a pump, probably no better, at best, than 20-30% in terms of compressed air energy to hydraulic output energy, and this is compounded by the fact that air compressors are also generally inefficient. Therefore the running costs of an air lift pump will be very high in energy terms. Second, it usually requires a borehole to be drilled considerably deeper than otherwise would be necessary in order to obtain enough [submergence, and this is generally a costly exercise. This problem is obviously less serious for low head applications where the extra depth [required would be small, or where a borehole needs to be drilled to a considerable depth below the static water level anyway to obtain sufficient inflow of water.

3.10    IMPULSE (WATER HAMMER) DEVICES

These devices apply the energy of falling water to lift a fraction of the flow to a higher level than the source. The principle they work by is to let the water from the source flow down a pipe and then to create sudden pressure rises by intermittently letting a valve in the pipe slam shut. This causes a "water hammer" effect which results in a sudden sharp rise in water pressure sufficient to carry a small proportion of the supply to a considerably higher level.

They therefore are applicable mainly in hilly regions in situations where there is a stream or river flowing quite steeply down a valley floor, and areas that could be irrigated which are above the level that can be commanded by small channels contoured to provide a gravity supply.

The only practical example of a pump using this principle is the hydraulic ram pump, or "hydram", which is in effect a combined water-powered prime mover and pump. The hydraulic ram pump is mechanically extremely simple, robust and ultra reliable. It can also be reasonably efficient. However in most cases the output is rather small (in the region of 1-3 litre/sec) and they are therefore best suited for irrigating small-holdings or single terrace fields, seedlings in nurseries, etc.

Hydraulic rams are described more fully in Section 4.9.? dealing with water powered pumping devices.

3.11    GRAVITY DEVICES

3.11.1    Syphons

Strictly speaking syphons are not water-lifting devices, since, after flowing through a syphon, water finishes at a lower level than if started. However syphons can lift water over obstructions at a higher level than the source and they are therefore potentially useful in irrigation. They also have a reputation for being troublesome, and their principles are often not well understood, so it. is worth giving them a brief review.

Fig. 77 A to C shows various syphon arrangements. Syphons are limited to lifts of about 5m at sea level for exactly the same reasons relating to suction lift for pumps. The main problem with syphons is that due to the low pressure at the uppermost point, air can come out of solution and form a bubble, which initially causes an obstruction and reduces the flow of water, and which can grow sufficiently to form an airlock which stops the flow. Therefore, the syphon pipe, which is entirely at a sub-atmospheric pressure, must be completely air-tight, Also, in general, the faster the flow, the lower the lift and the more perfect the joints, the less trouble there is likely to be with air locks.

Starting syphons off can also present problems. The simplest syphons can be short lengths of flexible plastic hose which may typically be used to irrigate a plot by carrying water from a conveyance channel over a low bund; it is well known that all that needs to be done is to fill the length of hose completely by submerging it in the channel and then one end can be covered by hand usually and lifted over the bund, to allow syphoning to start. Obviously, with bigger syphons, which are often needed when there is an obstruction which cannot easily be bored through or removed, or where there is a risk of leakage from a dam or earth bund if a pipe is buried in it, simple techniques like this cannot be used.

Fig. 77 Syphon arrangements

In Fig. 77 A, a non-return valve or foot-valve is provided on the intake side of the syphon, and an ordinary gate valve or other hand-valve at the discharge end. There is a tapping at the highest point of the syphon which can be isolated, again with a small hand valve. If the discharge hand valve is closed and the top valve opened, it is possible to fill the syphon completely with water; the filler valve is then closed, the discharge valve opened and syphoning will commence.

Diagram B is similar to A except that instead of filling the syphon with water to remove the air, a vacuum pump is provided which will draw out the air. Obviously this is done with the discharge valve closed. The vacuum pump can be a hand pump, or it could be a small industrial vacuum pump. Once the air is removed, the discharge valve can be opened to initiate syphoning.

Diagram C shows a so-called "reverse" syphon, used for example where a raised irrigation channel needs to cross a road. Reverse syphons operate at higher than atmospheric pressure and there is no theoretical limit to how deep they can go, other than that the pipes must withstand the hydrostatic pressure and that the outflow must be sufficiently lower than the inflow to produce the necessary hydraulic gradient to ensure gravity flow.

3.11.2    Qanats and Foggara

Qanats, as they are known in Farsi or Foggara (in Arabic), are "man-made springs" which bring water out to the surface above the local water table, but by using gravity. Like syphons they are not strictly water lifting devices, but they do offer an option in lieu of lifting water from a well or borehole in order to provide irrigation. They have been used successfully for 2 000 years or more in Iran, and for many centuries in Afghanistan, much of the Middle East and parts of North Africa.

Fig. 78 shows a cross-section through a qanat; it can be seen that the principle used exploits the fact that the water table commonly rises under higher ground. Therefore, it is possible to excavate a slightly upward sloping tunnel until it intercepts the water table under higher ground possibly at some distance from the area to be irrigated. It is exactly as if you could take a conventional tube well and gradually tip it over until the mouth was below the level of the water table, when, clearly water would flow out of it continuously and without any need for pumping.

Qanats are typically from one to as much as 50 kilometres long, {some of the longest are in Iran near Isfahan). They are excavated by sinking wells every 50 to 100m and then digging horizontally to join the bases of the wells, starting from the outflow point. Traditional techniques are used, involving the use of simple hand tools, combined with sophisticated surveying and tunnelling skills. Many decades are sometimes needed to construct a long qanat, but once completed they can supply water at little cost for centuries. The surface appearance of a qanat is distinctive, consisting of a row of low crater-like earth bunds (or sometimes a low brick wall) surrounding each well opening; this is to prevent flash floods from pouring down the well and washing the sides away. The outflow from a qanat usually runs into a cultivated oasis in the desert, resulting from the endless supply of water.

Fig. 78 Cross-section through a Qanat

Efforts have been made in Iran to mechanize qanat construction, but without great success, although in some cases qanats are combined with engine powered lift pumps in that the qanat carries water more or less horizontally from under a nearby hill possessing a raised water table to a point on level ground above the local water table but below the surface, where a cistern is formed in the ground. A diesel pump is then positioned on a ledge above the cistern to lift the water to the surface.

3.12    MATERIALS FOR WATER-LIFTING DEVICES

This is a complex technical subject if discussed fully, but. it is worth briefly setting out some of the advantages and disadvantages of different materials that are commonly used, as an aid to appraising the specification of different equipment.

Four main considerations apply for construction materials used for pumping water:

1. strength; stressed components need to be able to function over a long period of time without either failing through overload or, more likely, through fatigue;
2. corrosion resistance and general ability to coexist under wet conditions;
3. resistance to wear and abrasion is important for components that rub or slide or which are in contact with flowing water if any particulate matter is likely to be suspended in the water;
4. cost.

As in most branches of engineering, nature has not been kind enough to offer materials which simultaneously satisfy all these requirements completely; invariably compromises are necessary. The important point is to be aware of these and to judge whether they are the right compromises for the application of interest.

It is worth reviewing briefly the pros and cons of various different materials which feature frequently in pumps and water lifts; these are also summarised in Table 7.

Table 7 RELATIVE MERITS OF MATERIALS FOR PUMPS

Fig. 79 Animal-powered Chinese Liberation Pump mechanism uses steel components (for strength) to good effect (see also Fig. 96).

i. Ferrous Metals

Most ferrous, or iron based, materials are subject to corrosion problems, but to compensate, they are perhaps the most familiar low-cost "strong material" that is widely available. Generally speaking iron and steel are best suited for use in structural components where strength is important but a surface coating of rust will not cause serious problems.

Ordinary mild steel is one of the most susceptible to corrosion. Iron and steel castings, except where they have been machined, are generally partially protected by black-iron oxide which forms when the casting is still hot. There are several methods to protect steels from corrosion, including conventional paints, various modern corrosion inhibitors which chemically bond with the surface of the metal and inhibit corrosion, various forms of plating and metallic coatings such as zinc (galvanizing) and cadmium plating. Various steel chromium and nickel based alloys, the so-called stainless steels, are also resistant to oxidation and corrosion, but they are not cheap.

Stainless steels do make a useful alternative to brasses and bronzes, but they are very difficult to machine and to work and therefore most pump manufacturers prefer non-ferrous corrosion resistant alloys. One important application for stainless steel is as nuts and bolts in situations where mild steel nuts and bolts readily corrode; stainless steel nuts and bolts are expensive compared with mild steel ones, but cheap in terms of time saved in the field on items that regularly need to be dismantled in wet conditions for maintenance or replacement.

A primary mechanism for corrosion of steel in wet conditions is if the steel is in combination with nobler metals, (e.g. copper), and there is an electrical link between them while both metals are in contact with water. This can encourage what is known as electrolytic corrosion, especially if the water has a significant mineral content which will generally increase its conductivity.

Therefore, ferrous components ought to be well protected from corrosion and generally are best suited as structural items not having any "high quality" surfaces in contact with water. An example of a bad use for iron, where it sometimes is applied, is as cast iron pump cylinders. Here the internal surface will often keep in quite good condition so long as the pump is worked, but any lengthy period during which it is stopped a certain amount of oxidation will occur; even a microscopic outgrowth of iron oxide (rust) forming will quickly wear out piston seals once the pump is started again. Obviously any thin internal coating or plating of a pump cylinder is not likely to last long due to wear. However, cast steel centrifugal pump casings are often quite satisfactory, although parts requiring critical clearances such as wear rings are usually inserts made of a more appropriate corrosion resistant metal. Similarly, cast steel centrifugal pump impellers are sometimes used; they are not of the same quality as non-ferrous ones, but are obviously a lot cheaper. Pumps with steel impellers usually cannot have close clearances and machined surface finishes, so their efficiencies are likely to be lower.

ii. Non-ferrous Metals

Brass (a copper-zinc alloy) is commonly used for reciprocating pump cylinders. Due to its high cost, thin, seamless, brass tube is often used as a cylinder liner inside a steel casing, instead of a thick brass cylinder; obviously the steel must be kept from direct contact electrolytically with the water. Brass has good wear resistance in "rubbing" situations - i.e. with a leather piston seal, but it is not a particularly strong metal structurally, especially in tension. So called "Admiralty Brass" includes a few percent of tin, which greatly improves its corrosion resistance.

The bronzes and gun-metals are a large family of copper based alloys, which are generally expensive but effective in a wet environment; they usually have all the advantages of brass, but are structurally stronger, (and even more expensive too). Bronzes can contain copper alloyed with tin, plus some chromium or nickel in various grades and traces of other metals including manganese, iron and lead. So-called leaded bronzes replace some of the tin with lead to reduce costs, which still leaves them as a useful material for pump components. The inclusion of antimony, zinc and lead in various proportions produces the form of bronze known as gun-metal, which is a useful material for corrosion resistant stressed components. A bronze containing a trace of phosphorus, known as phosphor bronze, is an excellent material for plain bearings and thrust washers, if run with an oil film against a well-finished ferrous surface such as a machined shaft. Aluminium-bronze, which is cheaper but less corrosion resistant, replaces much or all of the scarce and expensive tin with aluminium.

Bronzes are generally among the best materials for making precision components that run in water and which need good tensile strength, such as pistons, valves, impellers, etc. Castings with a good finish can readily be obtained, and most bronzes machine very easily to give a precision surface.

Other materials, such as aluminium and the light alloys are generally not hard or wear-resistant enough for hydraulic duties, although by virtue of being very light they are sometimes used to make portable irrigation pipes; however they are not cheap as pipe material and can only be justified where the need to be able easily to move pipelines justifies the cost.

iii. Timber

Timbers exist in a very wide variety of types; their densities can range from around 500kg/m³ (or less) up to 1 300kg/m³. They also offer a very large variation in mechanical properties, workability, wear resistance and behaviour in wet conditions. Timber is of course also susceptible to damage by insects, fungus or fire.

The most durable timbers are generally tropical hardwoods such as Greenheart, Iroko, Jarrah, Opepe, Teak and Wallaba. The durability of many timbers can be improved by treating them with various types of preservative; the most effective treatments involve pressure impregnation with either tar or water-based preservatives.

One of the main factors affecting the strength of a wooden member is whether knots are present at or near places of high stress. Where wood is used for stressed components, such as pump rods for windpumps or handpumps, it is important that it is finegrained and knot-free to avoid the risk of failures. Good quality hardwoods like this are not easily obtained in some countries and, where available, they are usually expensive. Cheap wood is limited in its usefulness and must be used for non-critical components.

Certain woods like 1ignum-vitae have also been used in the past as an excellent plain bearing material when oil lubricated running against a steel shaft, although various synthetic bearing materials are now more readily available and less expensive.

Wood is available processed into plywood and chipboards; with these a major consideration is the nature of the resins or adhesives used to bond the wood. Most are bonded with urea-based adhesives which are not adequately water resistant and are not suitable for outside use, but those bonded with phenolic resins may be suitable if applied correctly and adequately protected from water with paints. Therefore, for any irrigation device it is essential that nothing but "marine" quality plys and chipboards are used.

iv. Plastics

There is a large and growing family of plastics, which broadly include three main categories; thermoplastics, which soften with heat (and which can therefore readily be heated and worked, moulded or extruded); thermosets (which are heated once to form them during which an irreversible chemical process takes place) and finally various catalytically-cured resins. As with almost everything else, the better quality ones are more expensive. Great improvements are continuously being introduced and there are interesting composite plastics which contain a filler or a matrix of some other material to enhance their properties at no great cost.

Although plastics are weaker and softer than metals, they generally nave the virtue of being compatible with water (corrosion is not a problem) and although their raw materials are not always low in cost, they do offer the possibility of low cost mass-production of pipe or components.

Thermoplastics based on polymerized petro-chemicals are generally the cheapest plastics; those used in the irrigation context include:

PVC (polyvinylchloride) is commonly used for extruded pipes; it can be rigid or plasticized (flexible); it is important to note that only certain grades of PVC (and other plastics) are suitable for pipes to convey drinking water for people or livestock, since traces of toxic plasticiser can be present in the water passed through some grades. PVC is relatively cheap and durable, but it is subject to attack by the UV (ultra-violet) wavelengths in sunlight and should therefore either be buried to protect it from the sun, or painted with a suitable finish to prevent penetration by UV radiation. PVC is also a thermoplastic and therefore softens significantly if heated above about 80°C; however this is not normally a problem in "wet" applications.

High density "polythene" (polyethylene) is cheaper and less brittle than PVC (especially at low temperatures) and is commonly used to make black flexible hose of use for irrigation, but it is also structurally much weaker than PVC, which is not necessarily a disadvantage for surface water conveyance at low pressure; however PVC is better for pressurized pipes.

Polypropylene is in the same family as polythene, but is intermediate in some respects in its properties between polythene and PVC. Polypropylene is less liable to fracture or to be sub-standard, due to bad management of the extrusion equipment, than is PVC; i.e. quality control is less stringent, so it can be more consistently reliable than poorly produced PVC.

None of the above plastics are generally applicable for manufacturing pump components for which strength and durability are important; these require more expensive and specialized plastics, such as nylons, polyacetals and polycarbonates. Nylons can be filled with glass, (for strength), molybdenum disulphide (for low friction), etc. An expensive specialized plastic of great value for bearings and rubbing surfaces on account of its low friction and good wear resistance is PTFE (polytetrafluoroethylene); certain water lubricated plain bearings rely on a thin layer of PTFE for their rubbing surface, and this proves both low in cost and extremely effective.

There are also various specialized thermoset plastics which find applications as pump components; these tend to be tougher, more wear resistant and more heat resistant than thermoplastics, and therefore are sometimes used as bearings, pump impellers or for pump casings. They are also useful for electrical components which may get hot. Most "pure" plastics are inclined to creep if permanently loaded; i.e. they gradually deform over a long period of time; this can be avoided and considerable extra strength can be gained, through the use of composite materials where glass fibre mat (for example) is moulded into a plastic. Various polyesters and epoxides are commonly used to make glass-reinforced plastics (g.r.p. or "fibreglass"); these are used to make small tough components or, in some cases, to make large tanks. Another example of composite plastics is the phenolic composites where cloth and phenolic resin are combined to make a very tough and wear resistant, but readily machinable material which makes an excellent (but expensive) water lubricated bearing, such as "Tufnol".

3.13    SUMMARY REVIEW OF WATER LIFTING DEVICES

Table 5, which introduced this chapter, is sorted into categories of pump types based on their working principles, but it is difficult to see any pattern when looking through it. Therefore, Table 8, which concludes this chapter attempts to quantify the characteristics of all pumps and water lifts in terms of their operating heads, power requirements, output and efficiency. Finally, Fig. 80 (A, B and C) indicates the different categories of pump and water lift demarcated on a log-log head-discharge graph (similar to that of Fig. 11). Obviously there are no hard and fast boundaries which dictate the choice of pump, but the figure gives a graphic indication of which pumps fit where in terms of head and flow, and hence of power. Note that Table 8 shows input power requirements, whereas Fig. 80 gives the hydraulic power produced, which will be a lesser figure by the factor of the pump efficiency. Due to the use of the log-log scales, the smaller devices appear to occupy a larger area then they would if linear scales had been chosen, however in this case it would not have been possible to fit sufficient detail in to the corner where the multiplicity of low-powered, low-head and low-flow devices fit, had a linear scale been applied.

Table 8 REVIEW OF PUMPS AND WATER LIFTS

Fig. 80a Typical head and discharge capacities for different types of pumps and water-lifting devices (on a log-log scale) (and continued in Figures 80b and 80c)

Fig. 80b DISCHARGE

Fig. 80c DISCHARGE