Introduction: How to Make a Radial Piston Water Pump
For a project ("Hydro do that?") at the Glasgow School of Art, our group of students was tasked with designing and manufacturing a water pump. The task was to be a competition betwixt several teams, each taking the name of an aquatic creature. Our designated company was 'Dolphin'. The winners of the competition were to be the group whose pump had the highest efficiency, and the pumps had to be able to pump five liters of water, through a height of 60cm (2 feet) in less that five minutes.
Our research and development led us to find that a piston pump would be the most effective way of pumping water.
We made three pump prototypes, to see which would be most effective at pumping water.
We built an axial impeller, a diaphragm pump and a piston pump.
The piston pump seemed to be the most viable solution to the efficiency problem, and our online research agreed with this, so we decided to build our final design in the form of a piston pump.
However, our brief required that we do testing and optimization for the pump, so we designed our pump to have three removable and interchangeable cylinders, to allow us to test the efficiency for different numbers of cylinders.
2mm aluminium sheet
3mm acrylic sheet
4mm mild steel rod
24V, 5600 RPM electric motor
Clear plastic pipe (unknown origin)
PVC pipe T-Joints
PVC plumbing pipe
Evian bottle necks and caps
Assorted screws, bolts, washers
Plastic Weld (Dichloromethane)
Band saw (metal)
Band saw (wood)
Sandpaper (various grit)
Assorted drill bits (including forstner / flat bits)
hot glue gun
Yeah, we used a lot of stuff on this project. The art school has a very well equipped workshop. One could probably substitute our pistons for something a bit more accessible, perhaps plastic and a standardized pipe size in order to make the whole task a bit less demanding in terms of equipment.
Step 1: Construct the Crank and Valves
We started the build by constructing the master/slave crank rod assembly. This was made out of aluminium sheet. Two slave rods were cut, a master rod (which was essentially the slave rod, but with a larger circle on one end of it), and a spacer plate. The spacer plate and the circle were to form the central hub of the crank arrangement. Each part had a 12mm hole drilled in the middle, and six smaller holes drilled around it, equally spaced to take the bolts that held the slave rods, and the whole assembly together.
The plates were spaced out by having an extra nut on the bolts in between the two plates.
Attaching the crank shaft was a bit of a fouter, as the crank had to be bent out of 4mm mild steel rod, but it had to be bent with the crank rod assembly already on it.
It was possible to bend one side first, then slide the crank rod assembly onto the shaft. Through use of a steel box-section jig, we were able to bend the other half of the crank and make it line up. The stroke on the pistons was 50mm, so the crank had to be offset by 25mm.
The two-part flapper valves were designed to fit the bottle caps we were using as the pipe attachment system. Since we needed all the valves to be the same, we had them laser cut from 3mm acrylic, and 0.5mm styrene.
We took measurements from the PVC T-joints, to make sure that we would laser cut the adapter pieces with the right inside diameter.
The valves worked by having two hinged semicircular flapper plated, hinged across the diameter of the pipe. The original valves had to be replaced because of a glitch whereby the two plated (sharing the same hinge hole) would overlap. This was solved by placing a strip between them, and stopping them overlapping.
Step 2: Pistons and Cylinders
The pistons were machined out of aluminium bar that just so happened to be the correct size for our cylinders.
They had a groove put in them 5mm from the front end, to take the rubber O-ring that would seal the piston. As well as this, they had a large hole bored in the opposite end to take the piston rod, and gudgeon pin (wrist pin).
These were machined by putting a large-ish section of bar in the lathe, facing the end off, boring the hole for the piston rod, then using a parting off tool to cut the O-ring groove, and then finally parting off the finished piston from the section of bar. This was done thrice to produce the piston heads that were required.
These piston heads were taken to a pillar drill, where the hole for the 4mm gudgeon pin was marked and drilled.
The clear plastic we used for the cylinders was scavenged, and as such we can't advise you on where to get it. You could substitute with acrylic tube.
It so happened that the tube we used fit perfectly over the PVC T-joints, so we used pipe cement to stick them on and provide an airtight seal.
Step 3: Mounting Board and Hoses
This was cut out of 12mm ply. The center was marked to allow for the full 50mm stroke of the pistons plus the length of the piston rod. Six holes were marked out where the T-joint on the cylinder assembly would pass through the baseboard. These were cut with a forstner bit, although a flat bit would work just as well.
A circular piece of ply was also cut to hold the crankshaft and support the gearbox. This piece and the baseboard were marked to have 6 holes cut in them, which had long bolts going through them, with PVC pipe spacers.
It might have been an idea to use thinner pipe, as it turned out that the crank system hit off them ever so slightly (this was easily fixed by filing the piston rods slightly smaller in places).
It was important to note that the heads of the bolts were on the underside of the baseboard, as the addition of the gearbox later would make them inaccessible, and would make it impossible to convert the rig to 2 cylinder operation.
Three guides were made with brackets to hold the cylinders straight. The brackets were strips of aluminium screwed onto the guide, and the guides were just bolted to the baseboard to be removable.
These were simply hose section hot glued into holes cut into the bottle tops.
The holes were cut first, then the hose inserted, then hot glue added.
Step 4: Gearing and Drive System
The gearbox was originally at a ratio of 100:1, providing 10Nm of torque based on the motors torque rating (it was later changed, as the device wasn't running fast enough).
The gears were designed in a gear generation program, and ported into Adobe Illustrator, in which we also created the pattern for each gearbox layer. We laser cut the pattern and double layered the gears so that they could take higher torque than a single 3mm layer of acrylic. To attach them to the 4mm mild steel shafts, we drilled a 1.5mm hole in the shaft, and put a pin through it and the gear, which allowed them to sustain high torque. The motor gear was simply push fit onto the motor shaft.
The motor was screwed onto the gearbox with two screws, and the whole thing was glued together with plastic weld, or dichloromethane.
Remember not to get your fingers caught in a gear system, as 10Nm of torque really hurts.
As an afterthought, we added a flywheel to our pump, to smooth out the reciprocating motion and increase efficiency.
This was made by taking a circular piece of plywood, marking out six points on the edge, boring large holes halfway through, and drilling small holes all the way through.
Large (15mm thick) steel slugs cut from wide steeel bar were bolted on into the holes. This will give the flywheel a large moment of inertia and smooth out the motion of the pump. Be careful with these flywheels, as storing a lot of energy in rotation can go wrong if there is a jam in the machine, or it suddenly stops.
Under no circumstances try to stop a flywheel with your hand. One of our group members cut their hand trying this, and trust us, it's not a fun thing.
Step 5: You're Done! Go Pump Some Water!
We thoroughly enjoyed building this water pump, and we hope you've enjoyed this tutorial, brought to you by Dolphin Dynamics.
Sadly, we did not win the competition. First place went to a rather well engineered elephant (rope) pump, and second went to another lovely pump, a diaphragm pump. the maximum efficiency of the winning pump was around 12%, ours only managed somewhere in the region of 6-7%. This was enough to net us third place in the competition, as well as a further commendation for audacity in design.
Some of the other pumps included another (single) piston pump running off a meccano gearbox, another rope pump, a reciprocating flap valve pump.
If you're interested, for the last two times this competition has been run, the elephant (rope) pumps have been by far the most efficient pumps out of the various different designs that have been seen.
Here you can see some of the prototypes we worked on before arriving at the final product.