Introduction: How to Make a Rotary Powered Pump

Hello, we are a team of 3 Product Design Engineering students at the Glasgow School of Art. For a second year project called "Hydro Do That?" we were tasked with pumping 5 litres of water up a height of 60cmin the most efficient manner possible. The pump had to be capable of performing this task in under 5 minutes, powered by a 24V DC motor . This project was a competition between several groups formed by students in our year. Team names were derived from aquatic creatures, hence we were known as Team Tortoise.

After several sessions of brainstorming and concept generation we had multiple designs in mind, including a centrifugal pump and a reciprocating valve-operated design. After some prototyping and testing we selected the reciprocating pump design as it was simple, effective and (we hoped!) efficient.

Here is how we constructed our design (and some of the challenges we met along the way.) Any feedback you can give us would be appreciated so feel free to contribute :)





Plastic tubing

Rubber seals (same diameter as the tubing)

Steel rod

Steel plate

Aluminium rod

Thick acrylic plastic

Coarse sandpaper


Squash ball

Also Used:

3D printed components


As you can see we used quite an extensive range of materials, however the main issue for anyone trying to recreate this will likely to be the equipment you have access to. We were using an extremely well stocked workshop, with a 3D printer to produce our guides, a laser cutter to cut our scotch yoke, an angle grinder to cut our steel components and a lathe for other components such as bushings.

Step 1: Step 1 - Yoke, Guides and Clamps

Here you can see the majority of our components, as well as close-ups of our guides and sliders.

The yoke was lasercut to ensure a precise form and smooth edges, reducing friction when in motion.

The guides were 3D printed for accuracy, as they are a relatively complex shape and required quite high tolerances to avoid jamming. They also have low friction when in contact with the plastic yoke.

The clamps were also 3D printed to ensure a good fit with the cylindrical pipe.

Note the sandpaper glued to the inside of the curved clamps: this prevents the cylinder from slipping once filled with water, as it gets quite heavy.

Polished metal contacts are fitted inside of the guides - these touch the yoke in operation, and also act as spacers, keeping it off the backboard.

Step 2: Step 2 - Axle and Rotary Arm

The next step is to make the central axle, which is connected to our drive system. The axle turns an arm, which is attached to the scotch yoke. The rotary motion of this arm produces the linear up-down oscillation that displaces fluid in our pump. We used metal components as they were required to withstand considerable torque.

The axle is made of mild steel rod. It is threaded at one end, and has been cut into a polygonal shape at the other - this allowed us to attach a drill chuck and rotate the axle without slippage.

The arm is also made of mild steel, and has 2 threaded holes in it. One hole mates to the threads on the end of the axle. We used a regular right-hand thread, which meant that the axle could only be rotated clockwise, otherwise the arm would unscrew itself from the axle. The other hole is countersunk and is threaded for a smaller screw.

Next, a small bolt was inserted tightly into the countersunk hole, facing in the opposite direction from the axle. (See photo) The screw must fit flush into the arm or there will be clearance issues!

The last photo shows the assembly fitted into the MDF backing. The axle has been inserted through a bushing, and the arm is resting on the yoke.

Step 3: Step 3 - Bushings

We made 2 bushings from aluminium, using a lathe to shape them. One was press-fit into the backing board. The precise shape ensured our axle remained in place, and created less friction that a hole simply drilled into the MDF backboard. The other bushing was smaller and was threaded inside. It fitted onto the small screw at the end of the rotary arm. It has a small lip machined into it, which is larger than the inside diameter of the scotch yoke channel, preventing it from disengaging. The photographs show the large bushing fitted into the backboard (with axle and the top-hat bushing connecting the rotary arm and scotch yoke.

Step 4: Step 4 - Backboard

We made a simple support to hold the pump vertical. It consists of 2 MDF sheets at a right angle, forming a base and a mounting surface. A large hole was drilled in the centre of the mounting surface using a fostner bit. Holes for the bolts the hold the guides in place were also drilled. In order to improve appearances we then spraypainted this board white.

NB: Provided guide mounts and central axle bushing are accommodated there is a lot of flexibility with the shape of this item. It can easily be customised to your own requirements!

Step 5: Step 5 - Pipe and Valve

The most complex part of our design is providing the linear motion to power it. The pump itself is simply a tube with a valve at the base. The valve allows water to enter the tube on the down-stroke, but prevents it leaking out on the up-stroke.

For the tube we used some simple plastic piping. Part of our brief requires experimentation to determine the best efficiency. When considering pipe diameter, a larger diameter can carry more water per stroke, however this weighs more, requiring increased motor power and causing increased stress to the rotary mechanism. Too small a tube diameter will not allow much water to be pumped per stroke - however the mechanism can be run faster. We decided that the diameter shown in the pictures was the best compromise available to us.

After some vigorous debate we settled on using a ball valve instead of a flapper valve. Our ball valve was constructed from a chamber of larger diameter piping, with a weighted squash ball inside. On the down-stroke water pressure forced the ball upwards, allowing water to enter the tube, but on the return stroke the ball sealed the inlet hole, sealing the water inside the pipe.

A cross-piece of wire was used to prevent the ball being pushed up too high and sealing the top of the chamber.

As the plastic of our tubing would not take glue we had to use a soldering iron to heat-seal the squash ball inside the 2 halves of the chamber.

Our weighted squash ball was made by cutting one open, filling it with iron filings and glueing it closed again.

We had several issues with this valve design, which is why you can see it reinforced using zipties in some photographs. More on this later!

Step 6: Step 6 - Assembly!

The final step is to assemble all of the individual components:

First, insert the axle and rotary arm through the central bushing.

Then the scotch yoke is fitted over the rotary arm - held in place by the top-hat bushing

Next, position the sliders over their mounting holes - do not bolt in place yet!

The guides fit over the scotch yoke and are bolted to the backboard - the bolts also hold the sliders in position.

Then the bottom clamps are held in place through bolts in the yoke. The pipe is positioned to the correct depth and then the upper clamps are bolted down, holding everything in place.

The final assembly step is to screw the ball valve on the bottom of the pipe.

To operate the pump the axle should be turned clockwise - pump speed can be adjusted by varying the speed of the axle.

Step 7: Operation!

Step 8: Issues

Unfortunately, we encountered multiple issues during this project.

As you saw in the previous videos, our pump is being directly driven, rather than using the provided 24V DC motor. We spent considerable time designing a gearing ratio and interface in order to utilise this motor, however when it came to operation it failed. After much frustration and many last-ditch attempts we still could not get the pump to function using the motor by the day of the competition. We believe that our design requires considerable speed to operate - the down-stroke must be powerful enough to force water into the tube, and it simply was not possible to attain this using the provided motor whilst still having enough torque to lift the system as it filled with water. However, as a concept our pump did prove successful, and when directly driven using a drill was one of the fastest pumps, successfully lifting 5 litres in about 15 seconds. Our pump also proved extremely robust, and could be directly driven at impressive speed when using a more powerful drill - however this led to a secondary issue involving the valve becoming apparent.

The ball valve proved acceptable at lower powered testing, but once a more powerful drill was used we found that it actually split apart under pressure. The fact it had to be heat-bonded rather than glued was probably a primary cause of this, and in our repairs we added zipties to help hold it together. The cross-piece that prevented the ball rising too much also proved delicate under higher power applications. In retrospect, a larger ball valve may have been stronger, and we would have been able to use a larger, stronger cross-piece. In our small valve there simply was not enough space to insert a strong barrier to prevent the ball rising too much.