top of page

Pedal Power Malawi

Clearly always a cyclist at heart, at university as part of a small team of mechanical engineers we set about designing and building an efficient pedal-powered irrigation pump to aid the lives of people in Malawi. This formed part of the university’s wider Malawi Project aimed at improving lives in Malawi as part of Scotland’s Millennium goals.

This blog captures that project in detail and hopefully shows the power of cycling and engineering combined. As a STEM ambassador, encouraging more people to do engineering in one of my broad goals in life, particularly since there is a shortage of engineers.


This group project focuses on providing an irrigation system for a crop farm in the small village of Namila, located in the third world country of Malawi, as part of Strathclyde University’s ongoing Millennium Project. A product design specification for the irrigation system was produced after consulting with relevant industrial and academic figures with experience of similar past engineering-based projects in Malawi. The group’s task in hand was to provide an efficient and feasible pump irrigation design solution to provide 30m³ per day to a garden adjacent to a tributary of the Shire River. The design was then manufactured and tested to determine its suitability for use and the potential for the design to be implemented throughout Malawi.

Transmission System

Belt drive

+ Small modification of bike required, low cost relative to chain drive

- Moderate friction losses, belt slip, high surface/air temperature resulting in possible expansion of belt, belt wear

Chain drive

+ Around 98% efficient, reliable, Long-lasting, readily available, high torque at low speed.

- Relatively complex, costly bike modification needed, potential corrosion of secondary chain

Rolling contact

+ Minimal modification of bike required, low cost

- High friction losses, tyre/roller wear, tyre failure risk due to combination of high surface and air temperatures and contact pressure, high stability required to maintain adequate contact

Fundamental Fluid Mechanics of Reciprocating Pumps

A reciprocating pump is essentially a piston moving to and fro within a cylinder. The piston is driven by a crank which in turn is powered by a prime mover. When the piston moves up within the cylinder the pressure is reduced in the cylinder which enables atmospheric pressure acting on the free surface of the lower reservoir to force the liquid up the suction pipe and into the cylinder. On the suction side of the pump there is a one way valve which opens, allowing water into the cylinder. This stroke of the piston constitutes the suction stroke. When the piston reaches top dead centre the suction stroke is followed by a delivery stroke in which the water is pushed out of the cylinder through another one way valve.

After a thorough study of the theory of piston pumps, it was decided that the pump would be double acting, i.e. there would be two pump cylinders in parallel acting 180° out of phase. Not only does this increase the volume of water pumped but also reduces the total friction loss due to inertia by maintaining a more steady flow from the pump. A comparison of the flow fluctuations of single-acting and double-acting pumps is shown below.

Analysis using fluid mechanics was performed so that the optimal dimensions of the pump could be found.

The volume delivered in one stroke of the piston is simply the volume of fluid displaced by the piston and is given by the equation below.

The value obtained from this equation is doubled as there are two cylinders being utilised. The volume flowrate of the pump is given by the equation below.

Knowing the rotational speed of the crankshaft meant that the volume flowrate of the pump could be determined. It was found, that a human could pedal a bike at 140W for approximately two and a half hours and that the average pedal speed was between 50rpm and 60rpm. This information was used as a benchmark for the pump system. Knowing the volume flowrate allowed an estimation of the total volume that could be pumped by an operator in two and a half hours.

The above analysis is very simple due to the fact that frictional effects and losses are not taken into account. The main loss however is due to the inertia of the liquid in the pipes, commonly known as water hammer. This inertia is a consequence of the liquid in the pipes decelerating and accelerating after each stroke, resulting in additional pressures due to the incompressibility of water. A simple method to curb such inertia forces is to install an air chamber on the delivery side of the pump and the following calculations assume that such a device is present and that the inertia forces are thus ignored. To find the frictional losses in the pipe system the velocity of the fluid was determined from the equation below.

The Reynolds number was then found from the equation below.

Using the Reynolds number calculated above, the flow in the system’s pipes could be determined as being either laminar or turbulent, thus allowing an estimation of the friction factor. Laminar flow is defined as flows with Reynolds numbers less than 2000; above Reynolds numbers of 2000 the flow is deemed to be turbulent. For the purpose of analysis it proved more efficient to use empirical relationships for the friction factors, rather than using a Moody chart, as the empirical relationships could be easily incorporated into a spreadsheet. The empirical relationships for the friction factors of both laminar and turbulent flows are given in the following equations.

The head loss for turbulent flow in a pipe is given by the Darcy equation.

Sharp bends in pipes may cause the flow to separate from the wall, forming a restriction and an uncontrolled expansion further downstream. The same effect is evident at pipe entries from a large reservoir. The entry losses at pipe inlets, outlets and bends are given by the following equation.

In the above equation, K is known as the fitting loss coefficient. It is a non-dimensional constant and its value is found through experimentation. In this case the K value was estimated to be 0.5.

Knowing the total losses in the system allows the total equivalent head to be found using the following.

The input power required to drive the pump is then found from equation below.

If an overall pump efficiency (approximated at 90%) was assumed then the power required from the prime mover could be calculated. If this was determined to be over 140W then the design was not suitable based on the power that a human on a bike can generate.

It is important that the phenomenon of separation is avoided when designing reciprocating pumps as it is an important limitation on the location of the pump, in relation to the level of the lower reservoir. The manometric suction head represents the lowest pressure in the system (at pump inlet) and if this pressure falls to the value of the liquid vapour pressure, cavitation will occur and delivery will cease. On a basic level cavitation can be described as a local vaporisation of a liquid. In practice, cavitation occurs at pressures that are somewhat higher than the vapour pressure of the liquid. The main effects of cavitation are erosion (due to the formation of small bubbles) and vibration and as such performance failure.

Therefore in the case of the pump design it was vital to ensure that the absolute pump pressure did not fall below 2.4m absolute. To avoid separation the maximum permissible drive speed of the pump was found from below.

Transmission Analysis

Having generated concepts and carried out a qualitative analysis, it was then necessary to assess the performance of the selected potential transmission system designs using a theoretical approach.

Human Power Input

As the source of power for the pump drive system was to be human, the quantity of power which could be produced by a human over a sustained period would be the limiting factor on how much work the system could do.

The following graph - produced by Douglas Malewicki (1983, International Human Powered Vehicle Association Scientific Symposium) - illustrates the length of time over which a human can produce a set power output.

For the analysis it was assumed that it would be reasonable for a single person to cycle for two and a half hours. The pump output could be increased with multiple users cycling for shorter periods of time. However, it was assumed that there would only be a limited number of persons available to operate the pump, thus one user cycling for a prolonged period was deemed to be the best representation of the actual scenario. Therefore, from Figure 12, the sustained level of power which could be produced over this period would be 0.205hp based on pedalling for 2.5 hours.

With a conversion factor of 1hp = 745.70 W, the average human power output for a cycling duration of 2.5 hours would be 152.87W

For the single chain transmission a transmission of efficiency, ηtr of 97% was assumed. Therefore, the shaft power available at the pump crank, Pin would be 148.28W.

This power could be related to a comfortable cycling speed via the graph shown below.

From the graph it can be observed that for a power output of approximately 150W a comfortable pedalling speed would be around 50 rpm.

It should be noted that the above power calculations were based on an average healthy human. These values may require adjustment dependent upon the user’s calorific intake and climatic conditions and are liable to fluctuate also from person to person.

Final Pump Design

The values of 125mm for the stroke length and 70mm for the cylinder inside diameter were the main constraints for the design of the pump. Due to the prescribed cylinder diameter it was possible to calculate the length of the piston using the equation below to give a value of 105mm. This is an industry standard formula, used for the calculation of piston lengths.

The wall thickness of the cylinder was investigated, in order to discover if there was an optimum thickness for the designated pressure inside the cylinder for the PVC material which was chosen for the cylinder material.

The calculation revealed that the optimum thickness for the cylinder wall was 0.0178mm. This value was deemed impractical as durability was a concern and as a result, a thickness of 5mm was chosen as the minimum practical thickness.

Several design concepts were derived for the design of the cylinder which can be seen below.

The design shown in the first image was discarded due to the difficulties attaching the cylinder to the base plate whilst still retaining a good seal. The principle of the design in the seconf image was to have an inner cylinder that was sealed against the wall of the outer cylinder using o-rings, and it would have inlet and outlet non-return valves contained within the inner cylinder. This design was discussed with the University Lab technicians and it was decided that the mounting of the valves internally would be time consuming and difficult to achieve. For this reason the second cylinder design concept was discarded. The final cylinder design is shown in the last image. This consisted of a one piece cylinder with a single opening in the bottom of the cylinder supplying both the inlet and outlet lines for the pump system with the non return valves placed externally of the cylinder. This design was chosen as it was considered to be effective and relatively simple to manufacture.

As the intended method of powering the pump was by bicycle, the rotary motion of the bicycle sprocket would have to be transferred to the pump. The decision was taken to base the transmission of the bike power to the pump loosely on that of an internal combustion engine, where the bike would drive a crank shaft, with a connecting rod between the piston and the crank. The piston design is shown below.

Another consideration was how to provide an effective seal between the piston and cylinder walls. From the research it was discovered that a common method of providing a seal in treadle pumps was to use leather. However it was also possible to use rubber o-rings and it was these o-rings which were used in this case.

As can be seen, it was necessary to have a pin which would pass through the bearing of the connecting rod, thus attaching the piston and connecting rod, whilst allowing free movement. There was some concern as to the shear stress which the pin would experience and as a result calculations were carried out to find the shear stress experienced by various pin diameters. The shear stresses were calculated with an approximated applied force of 600N.

Due to the small internal diameter of the cylinder, the connecting rod would have to be fairly small in size. The first connecting rod design can be viewed below.

The idea was to have the top and bottom ends of the connecting rod detachable so as to allow the fitting of the bearing into the groove seat. However, after discussion with the University technicians, it was decided that the connecting rod should be one piece with the seat for the bearing drilled into the top and bottom ends for an exact fit for the bearing. The connecting rod could then be heated in an oil bath to allow it to expand, and then the bearing could be pressed into position. This design can be seen the figure below.

A major problem due to the small diameter and the long stroke length was that during the stroke the connecting rod was at such an angle that it would strike the cylinder walls. The decision was taken to trim the height of the cylinder, and therefore when the piston was in its top dead centre position there would be a portion of the piston outside the top of the cylinder. While this solution was not ideal, it was considered to be the best option available.

A key component of the pump system was the crank which is shown below in the initial design concept of the crank. After consideration however, it was decided that this design contained too many flaws. Firstly, it was unlikely to be able to resist the torsional effects which would be present from the rotational motion. Additionally the resistance caused by the compression of the fluid inside the cylinder could have posed problems. Secondly, there was concern of how to construct the crank shaft, whilst being able to place the bearings required in the correct positions.

The image below shows the final design of the crack to overcome these limitations.

Pump Assembly

The constituent pump parts were assembled on to a large wooden pallet to provide a flat surface and to keep the parts the required distance apart. It also meant that the pump was extremely easy to install as the parts simply had to be bolted into the marked holes in the pallet using 20mm bolts.

The idea of using a pallet was also deemed appropriate as it was assumed that such materials would be available in Malawi and would be required to ensure that the pump parts are level on uneven farmland.

Volume Flowrate

It was hoped that the testing could be carried out up to a maximum head of 10m, however due to limitations on the ability of the pump system – which are discussed in the conclusions - the head was only tested between 0 and 4m. The graph below displays the variation in volume flowrate over the range of heads which were tested.

It was clear that the volume flowrate decreased linearly as the head increased over the small range of heads analysed. This linear relationship was as expected because the flowrate is inversely proportional to the head.

Conclusion - Pump Performance

Upon the delayed completion of a working pump prototype full testing was carried out to determine the performance of the pump. The pump performed reasonably against the expectations. The testing of the pump demonstrated flowrates that matched the theory closely. By using a simple blockage at the pipe outlet it was very easy to create an appreciable jet from the pump which had a range of approximately 5m. The pump could therefore be easily developed to incorporate simple spray irrigation.

The flowrates obtained from the pump were thought to be reasonable by the group in terms of the irrigation need in Malawi. The ergonomics of using the bike to power the pump were also fairly good. The rider reported relative comfort and little straining achieving high flowrates up to a head of 6m. The unsteady flow characteristics of the piston pump manifested in the rider’s feet often slipping off the pedals at the top of pedal stroke.

The pump however was not able to pump beyond a head of 6m. This was due to the high human power input required and the high stresses on the pump system. A head of 10m was specified in the design specification, however the group felt from the start that this head would be unattainable and were dubious of the need for a 10m head.

There was significant slippage from one of the pump cylinders, especially at higher heads as can be seen below.

It was postulated that the main reason for the slippage was due to the uneven surface on the inside of the cylinder due to the previously stated difficulties in boring the PVC.

Having only one o-ring on the piston also contributed to the slippage due to movement in the cylinder.

Conclusion - Transmission Performance

The chain drive used for the transmission worked well when coupled with the pump. One problem was that the sprocket on the crankshaft and the sprocket on the bike were not an exact match which meant that the chain was liable to slight torsional movement. This sometimes caused the chain to come off the sprocket. With the sourcing of a more appropriate chain the group was confident that this problem could be eradicated.

Future Developments

The scale of this particular project means that there is substantial scope for future development. The group views this project as being evolutionary in that it will progress over the coming years. To this end the group has had several meetings with the second Malawi group, who are developing future projects within the department, to discuss the potential for the irrigation system.

The group have applied for funding to allow several members to travel to Malawi during summer 2008 to implement the pump design and to develop a more robust framework for designing irrigation systems. Should funding be approved by the IMechE, Alan Johnson and David Lavery are committed to a meaningful trip to Malawi and the group as a whole are supportive of meeting with next years students working on the Malawi project to offer advice and guidance.

On the technical side of the project there are many possible developments and improvements that could be made. With the basic pump system in place there is countless improvements that could be made both in terms of the pump itself and the prime mover. Based on the testing experience, specific improvements that could be made to the pump design are:

  • Improving the pump stand to be able to take higher loads and to be more robust.

  • Fitting a second o-ring to the top of the pistons to reduce the movement within the cylinder and thus reduce the slippage.

  • Improving the air chamber design to help reduce flow fluctuations. Adding an air chamber to the suction side as well as the delivery side will also decrease the flow fluctuations.

  • Make the bike stand sturdier.

  • Different transmissions systems and gearing could be considered.

  • A guide sprocket should be incorporated to ensure that the drive chain tension is ensured.

The group stress that the pump manufactured for this project should be considered as a working prototype and the group feels that there is great scope for improving on this design. As such the group recommends that this project should be developed by subsequent Malawi project groups to optimise the pump design. Should the group attain the aforementioned funding for a trip to Malawi then the above improvements will be made by the group. The group are also willing to assist subsequent Malawi groups involved with this project with any help we can provide as the group feel that the Malawi project is extremely worthwhile and an excellent way for students to use the engineering skills developed through university in a positive way.

The most obvious extension of the project would be an investigation into the specifics of the irrigation. This project has concentrated mainly on the pump and has not considered the delivery of water to the field. This in itself is a separate project and the group foresees that success in this would lead to a more valuable overall solution. Interdepartmental cooperation would be beneficial in that, for example a biologist could be consulted regarding the water requirements of different crops or civil engineers regarding the storage tank and water purification.

bottom of page