I was tired of constantly battling with a malfunctioning petrol pump only to get meagre amounts of water for our animals. So I built this:
And it actually works!
I should backtrack. Let me explain. You see, animals need water – a lot of water – and water needs to come from a source. Luckily for us, a river runs around part of our farm. However, that river is twenty-odd metres below the stock tanks (water reservoirs on a hill, which gravity-feed into troughs). This means that we need to transport water to that height by one way or another.
The previous owner was kind enough to leave his old pump and motor, which he had installed down by the riverside. It was getting old, however, and was temperamental at best. It would take us multiple trips down to the river to refill the gas tank so we could fill our reservoirs on the hill.
Pumps and motors have a heck ton of moving parts (read: breakable parts), and combustion engines are just… No. We do want to revive the haggard old beast someday, if only for a restoration project (despite their age, they actually still sell for around two-and-a-half grand). But as it started to fail we realised it would require hundreds of dollars in spare parts to fix ourselves. As you can see, it needed constant attention:
We needed something reliable for the summers ahead, and knew that we had to come up with a better solution. I’d always fantasised about building something to harness naturally available energy (naive idealist here), and since a raging river was literally right there, it seemed like a waste not to try. So, with absolutely zero knowledge or experience in such matters, and in the true spirit of the oblivious DIYer, I set out to construct a renewable energy alternative. Huzzah!
There was more trial and error than you can imagine before I reached the working product depicted at the beginning of this post. One whole year earlier, I started my first prototype, which was doomed to fail. It all began with the contemplative idea of using an old plastic barrel as a waterwheel…
First, I cut it in half and divided the circumference into equal-length fins, which I then cut with an angle-grinder.
I heated parts of the plastic up with a heat gun and then bent the fins open, using clamps to hold them in place as they cooled.
I then bolted threaded rod through a steel round bar and onto the wheel.
I built a wooden frame to hold the bearings, and viola, we have an axle.
We had another old water pump lying around which still held together, so I decided to try and see if I couldn’t power it with my wheel.
Whilst rummaging through the multitude of potentially useful junk we’d reserved from our post-purchase farm cleanup, I also found a large rusty belt wheel. Perfect! I had in mind that something like that was needed for stepping-up the torque in the axle (like a gear) to the smaller wheel on the old pump. (I may sound like I know what I’m talking about, but the astute among you will note that I clearly don’t.) I stripped the belt wheel of rust, primed it, and spray painted it black.
Then I constructed a really stupid rig. There’s a reason for the convoluted contraption you see below, but the details are boring. Basically, I intended the waterwheel to be manoeuvrable in case of a flood, so I built a rig where the heavy-ass pump was fixed to the axle support. Just… don’t worry.
It did spin, at least. And it looked kinda cool. Here it is – you can even hear the little pump oscillating.
But the damn pump alone weighed twenty kilos! I had my doubts it would last up there, and really, it was only put there because I couldn’t conceive of a better place where it could lift with the axle. Thankfully, other plans would ensure failure of my first attempt before I wasted more time on this. Which brings me to the next part of my misadventure: The riverside installation of Prototype One.
The original location for my waterwheel was next to a small waterfall. On the outset this seemed ideal, because I could set the waterwheel up to be a “breastshot”. A breastshot waterwheel is where the flow of water hits the side of the waterwheel, rather than the bottom. The falling water imparts greater energy to the waterwheel. First I had to clear a bunch of mud and debris which had washed up near the waterfall on the riverbank.
I then created forms for concrete pads which would be the support blocks for my contraption.
I used a masonry bit to drill into the bedrock, and I smacked in some big bolts (yes, wrong kind, I know) to hold the concrete in place.
I used mud to hold the form in place so the concrete pad would be level. The mud also served to stop the wet concrete from spilling out the gaps underneath the form, which sat on very uneven bedrock.
You can also see in the above picture some plastic bags filled with mud. This was the beginning of the end for Prototype One. I put the bags there to block water which was seeping through the mud wall which I had dug out. Eventually the loose soil would wash away, diverting the water from from waterwheel.
With the pads and supports set, I waited a couple of days for it to firm up.
During that time I made a wire fence to deter the wayward cows who cross the river from the neighbour’s paddock on the other side. Cows are arseholes. They destroy everything. I wasn’t taking any chances.
Then the exciting part came of seeing my waterwheel spin!
It didn’t. So I added metal fins!
At this point I had my doubts about the project. As you can see between those two photos, the mud wall had eroded, diverting part of the river, which meant there was less water flowing where I had situated my waterwheel. I couldn’t lower it any more, either, and the fins would sometimes grate against the unseen bedrock, too. The floppy unsupported fins wobbled about, dissipating a lot of the inertia. Nevertheless, it was spinning, however weakly, so it was worth trying the next step.
I bolted the pump to a temporary platform and connected the two wheels with a fitted belt.
Dingus was visiting at the time, so he held up a pipe, measuring the water volume over RPM whilst we tested various speeds and approaches.
Dingus is studying engineering, so he was eager to see if a mathematical brain could shed some light on what forces were at play. There were so many unknowable variables, however, that despite being an interesting experiment, success would only be determined by arduous trial and error.
Annnd nope. The pump required too fast of an RPM. We could spin the pump by hand and it would spurt out a little water, but the waterwheel just wasn’t getting enough juice to keep up. Even if the waterwheel itself was redesigned (which it later would be) or I implemented some gears, this location was no longer optimal due to the terrain. Here is the final picture of Prototype One, looking oh-so promising.
A little disillusioned, and a whole lot o’ discouraged, I chewed my lip for a few months. I thought about a hydraulic ram-pump instead, but the river snakes and drops too quickly, and the numbers I ran made it seem unfeasible with what we had to work with, such as the location of our underground pipe which led up to the stock tanks. (I may revisit the idea of a ram-pump sometime in the future, because a whole lot more can go wrong with a waterwheel.)
It wasn’t until I went for a walk along the riverside that another location farther downstream reignited my hope. I’d found a natural weir – a flowing passage of water carved into the bedrock, ideal for what I had in mind.
It’s difficult to see in the photo, but below that beige rock the water is channelled quite deep before it becomes turbulent – a promising new location to inset a breastshot waterwheel. And the terrain for the supports was fairly level, to boot. It would be tight working between two large horizontally-growing trees which hung over the river to either side, but after a few measurements I was optimistic. Thus began Prototype Two!
I was determined to build a proper-looking waterwheel this time. I 3D modelled an eighteen-sided segmented polygon in Sketchup (so I didn’t have to wrap my brain around the geometry) and then cut all my lengths from treated pine, making sure to treat the cut ends as well, since the whole thing would be constantly wet.
Through a surprisingly quick and satisfying process of joining angles and screwing them to the new fin arms, a delightful shape began to emerge.
Soon I had something that resembled a large polygonal cog. I trued it by using a bar clamp to ensure it was a circle, not an oval.
I bolted two hexagonal hubs comprised of treated ply to both sides, with holes cut out to allow for the axle.
I cut out some metal fins from galvanised scrap that was lying around, seemingly from an old council sign someone had thieved/salvaged.
These were then screwed onto the wood frame. I bent the edges so the water would be scooped, adding more energy.
Here it is looking all sexy like:
Once again, it was time to repeat the process of creating concrete slabs and supports. This time, I measured everything at least four times to be sure it would fit, and that I would have wiggle room.
I made a template of the size of the waterwheel, which was used to find the sweet spot.
And again, I drilled into the bedrock.
But this time, instead of large bolts (which seemed to break away from the previous concrete slabs, causing them to wobble), I opted for a bunch of long thick nails, which I drove into the holes and then bent over to create better surface area onto which the concrete could cling.
Some wire debris was tangled up in a nearby tree, so I cut it down and put it to use, reinforcing my nails. Pretty sure this is industry standard.
I levelled the supports, using the nearby trees and a wedge in the ground as braces.
Then it was time to transport this badboy!
At over one-and-a-half metres in diameter and weighing about forty kilos – not to mention bedecked in sharp metal fins – it was a trick to get it over two wire fences and down a steep bushy riverbank. Thankfully, Dingus had returned to lend some manpower!
And just like that, it slotted right in place, exactly where it was intended to go, with shuffle room to spare. Ahhh.
Here, have a delectable video:
I wasn’t happy with how much axle was unsupported between the bearing and the waterwheel, or how short the axle support was. With no one around to help me lift it, I resorted to securing it with ratchet tie-downs so I could remove the axle and reinstall a longer support.
Ah, much better (bearing and support goes right up to the waterwheel):
I was fairly confident with the waterwheel build, in its sturdiness, simplicity, and potential to gather energy. What I wasn’t confident in was the next part of my hair-brained scheme: To make, from scratch, a PVC piston pump. SHIT JUST GOT REAL.
Building a waterwheel is fairly straight-forward, there’s not actually that much to it. It’s a wheel on a stick. Utilising its energy and transferring that into a pump intended to deliver water twenty metres vertically over two hundred metres… now that’s where things get complicated. But I’m not an extensive blueprint kind of dude; I prefer to waste as much time, energy, and resources as possible. Not really, but I do like to just intuitively nut things out, sometimes to the aggravation of my more technically-minded love-muffin.
There were four components to this endeavour which I had a rough instinct for but knew nothing about and had no real plan for: A mechanism to convert rotational motion into linear motion, a water-tight piston unit to push and pull water, check valves to ensure water would flow only in one direction to maintain pressure, and a system of maintenance, such as a ball valve tap, priming port, and pressure release. My biggest hurdle of the four which consumed the most time was the creation of the piston unit, which would see numerous revisions.
PVC components in New Zealand aren’t as ubiquitous as in other countries, so there were slim pickings. PVC pressure pipe would be used for the housing, but I needed to craft a piston head – a form-fitted plug – that could travel forward and backward inside the pipe and maintain pressure without leaking. My first idea consisted of cutting a part of the pipe and then filling it with hot glue.
This worked surprisingly well. I cut the pipe away once the glue had cooled, revealing a cylindrical plug which perfectly fit the pipe’s internal diameter.
After a touch of sanding, I cut grooves using the table saw for o-rings, which would ensure a water-tight fit. The kerf of the blade was coincidentally the same width as the o-rings I had.
I glued the piston head onto a smaller diameter pipe which roughly fit inside the piston housing.
Whether this would work or not I was unsure (it would, but only briefly). Only by assembling all the components would the flaws in my design be revealed. And boy, were there many.
Next up was the troubling and finicky challenge of building a mechanism to convert rotational motion to linear. Yeah. I was actually quite confident (read: arrogant) that my initial idea would work (once again, it would, but only briefly, and for a different reason). A crank arm, of course! So this is what I built:
Next up was the aspect I was most unsure about. The check valves (or non-return valves). If you don’t know, check valves are little valves that open automatically under pressure to allow water to flow in one direction but not the other. This way water can be sucked up into the pump on the pull stroke, but can’t rush back out on the push stroke. This allows you to direct water.
One thing to understand about check valves: They are f-ing expensive. For the size of my setup, I would need to spend around a hundred bucks for two check valves. Nope. I was going to make my own. Out of rubber bouncy balls.
I’d read about people doing this for compressed-air/vacuum potato cannons and homemade bicycle pumps. If it works for air, surely it’d work for water, right? I thought it might be worth a try.
It’s difficult to explain, so if you don’t know, I made a diagram of my setup so you can get your learning on:
I drilled holes through the PVC couplings and put brass pins through to block the rubber balls (denoted by the black dots in the above diagram). Silicone was applied to prevent leaks. I oriented the couplings in such a way that the rubber balls could be replaced if they degraded over time. Making this setup serviceable was one of the biggest challenges.
The final component was the simplest of the four. It consisted of a bunch of polyethylene agricultural pipe connectors arranged in the right order.
It took a bit of fiddling to understand how to prime the lines and avoid air locks. (An air lock is where a pocket of air gets caught in the lines and the pressured water just compresses it like a cushion, preventing flow.) But I got there after a few grunts and curses. Here’s the service components once I installed them later on:
With this arrangement, it meant I could turn the line off with the ball valve, preventing all the water in the pipe from travelling backward, if I needed to work on the other components. The pressure release allows me to test adjustments to the other components before turning the line back on to full pressure. The priming port is so I can fill up the lines before the ball valve if I’ve been doing some maintenance, so an air lock is avoided. Hope that all makes sense.
Backing up a little, in this video I test the pump alone for the first time without the pipe to the tanks being attached (or the service components):
After this I connected the outlet pipe directly to the pump to see if the setup worked in its current state. CATASTROPHIC FAILURE. Well, maybe not catastrophic, but it almost broke my entire pump. As the piston was pushed by the crank against the pressure in the pipe, there was nothing bracing it, so it bent upwards. Thankfully the PVC was forgiving. Before trying again, I made a rail-type brace with a wheel attached to the piston so it could stay in line with the pump.
Here’s a video featuring the silly piston brace (which served its purpose before I changed the whole mechanism later on), and a pressure test after connecting the service taps/ports but before connecting the outlet pipe:
And in the following video it’s finally attached to the lines and working. Note how the waterwheel slows down on the push stroke of the piston as it works to push against the pressure in the pipe (about 21 PSI):
I waited for ten seconds, biting my nails, wincing, preparing for catastrophic failure… Twenty seconds… Thirty seconds… Minutes passed… But it held. IT HELD! I excitedly ran up the hill, jumped on the quad bike, and zoomed off to the stock tanks to see if they were filling…
NO WAY… It actually works?! After so many discouraging mishaps and never knowing if it would be possible, seeing this sight almost made me cry. I couldn’t believe it at first. I had so much doubt. But it worked. It actually worked! And had a decent output over a minute:
It might not look like much, but because this thing is powered by the river itself, it’s designed to work continuously (or at least until the tanks are full, then it can have a break). Our target is two thousand litres per day. This output guaranteed five thousand. More than double what we need. So even if I were to make some mechanical trade-offs down the line to improve efficiency but reduce output, I can decrease the rate of flow by more than half. Woot!
I left it on overnight, but came back to this:
Under the immense amount of torque, the whole axle had snapped at a weak point- a previous hole that had been drilled through it for Prototype One. Damn. This meant buying and reinstalling a whole new round bar. Other failures have happened since, too, such as the snapping of the bolt that holds the waterwheel onto the axle, resulting in my shock when I discover my waterwheel missing, presumably washed over the huge waterfall downstream. Luckily I found it caught up not far from the setup, submerged but still intact.
Whilst I was fitting a new axle, I wondered if there was too much resistance from the crank arm. Seeing it slow down on the push stroke did make me dubious about its durability. Also, would it be strong enough in summer when the water level was lower and there were longer periods without rain?
I was concerned. So for the first time during this whole project, I actually did some research. I looked into various ways of translating rotational motion into linear motion, and found that for high torque applications such as this, a “scotch yoke mechanism” is significantly more efficient than a simple crank arm. So, I upgraded. This is my hodge-podge scotch yoke:
When I was making the mechanism, I also chucked the original PVC piston shaft and hot glue piston head (which was leaking) and made a new piston head from a sanded dowel which I coated in polyurethane before attaching. Zero leaks.
The piston head was bolted onto some leftover round bar, which I connected to the scotch yoke. And here it all is in motion, with an added make-shift linear bearing comprised of small wheels to hold the round bar in line with the pump:
And here’s another angle:
As you can see, it’s very rudimentary at this stage. A definitive prototype. Many, many, many kinks to iron out. But guess what? IT WORKS. Here’s the whole unit again in all its glory (you can see with the scotch yoke mechanism the waterwheel barely slows at all on the push stroke):
Even if this whole assemblage breaks, it won’t have been for nothing – at least I’ve proven that it is possible and feasible to pump water up from the river to our stock tanks using a homemade waterwheel and homemade piston pump. Yus!
Now I just need to save up for a welder and learn how to weld so I can make a more robust and condensed version… Hopefully this one will last for the time being!
Update: Our waterwheel story continues here.