Over the summer just past we experienced our first bad drought here on the farm. Most of New Zealand was hit hard, followed up by a sucker-punch from COVID-19 that added insult to injury for already-struggling farmers. Along with cutting into our winter feed supply because the grass wasn’t growing, another major challenge we had to contend with on our farmlet was the malfunctioning water supply for our hot and thirsty livestock.

One of our ongoing DIY projects has been our waterwheel pump, whose early iterations I posted about a while back. Over the past two years, the (sometimes) working prototype has seen a variety of experimental amendments and re-imaginings. But we were forced to radically step up our game when the river’s water level started running worryingly low, refusing to turn the waterwheel at all…

Through toying with various pulley arrangements (learning how pulley ratios worked as we went), teaching ourselves to weld, figuring out and refurbishing a hundred-year old piston pump, and extensive face-palming with trial and error, we surprised ourselves by creating a system that supplied more water than our animals demanded even during a drought. Victory!
Before I delve into the harebrained and arduous journey of how we reached the final product, here’s a video tour of our current working waterwheel-pump setup:
So, now that you’ve seen the satisfying conclusion to our water supply tale, let me show you what it took this particular fool to get there.
Going back in time, perhaps a year or so ago, we were still using the PVC pump I had constructed, with a scotch-yoke mechanism converting the rotary motion to linear. This was far from efficient, with high levels of friction and wear. You could even see the drive shaft bow slightly!

(Bonus: In the above pic, Char inadvertently captured photographic proof of the Stonybrook monster bathing in the river…)
For a time, the scotch-yoke served its purpose, working well enough to supply water to the farm for – I think – over six months. During that time, I replaced the crappy wooden piston head I made with a proper one that I had custom machined at the local engineers. I designed it in Sketchup, with the intent to have a pivoting joint near the head for a future idea I was working on.

The new machined piston head (with channels for o-rings) was form-fitted to the PVC pipe I was using for the piston cylinder. Yay, no more leaks! (For now…)

Tight fit. It needed a good lubing.

Get your mind out of the gutter. Shame on you.
So anyway, the new form-fitted penis piston head solved some issues, but with the scotch-yoke mechanism seemingly degrading in efficiency, the waterwheel was slowing down considerably on the push stroke, especially when the river was low.
Since the waterwheel slowed on the push stroke (pushing water up twenty metres) and sped up on the pull stroke (pulling water up from only a metre or less), I tried to steal some of the excess energy of the pull stroke to add to the push stroke. First I tried a cable weight, which rose the weight on the pull stroke, converting that potential energy to the slower push stroke.
Understand? Don’t worry, it’s not important and didn’t amount to anything, just nod and smile.

After testing with a poorly balanced sack of gravel, I zip-tied some heavy steel miscellany to a upturned pulley, which weighed ten kilograms or so altogether, suspended by a cable that was attached to the scotch-yoke.
This strange setup did actually improve efficiency, so I made it semi-permanent by substituting the cable weight for a strong spring, which works on the same principle.

So as the piston shaft was pulled back, the spring would stretch. Then when the piston shaft was pushed forward (pressurising the water line uphill, which has more resistance), the elastic energy stored in the spring was transferred. It was working harder on the easier pull, and working easier on the harder push. A trade-off.

I also added some weights to the outer edge of waterwheel itself, at strategic locations that seemed to help it maintain momentum. I also put a pipe farther upstream at a higher elevation, with the end of the pipe positioned above the waterwheel fins, to add a bit more gravity to the equation.
Desperate, I even put plastic containers on certain fins, which were filled by the pipe I put upstream. It was messy, inelegant, and unappealing. It did, however, work… for a time.

But when the river’s water continued to subside, and our water demand wasn’t being met, I was forced to throw out the high-friction scotch-yoke mechanism. (At this point I should have experimented with widening the fins, but it wasn’t until Char chimed in later on that I realised I could.)
I instead utilised the pivot point I had designed into the piston head. This way, the piston shaft would pivot vertically as the piston head plunged into the piston cylinder (like a crank arm).

But this only lasted so long, as the PVC diameter was too small for the angling piston shaft. And things tend toward chaos over time, especially if you’ve bolted everything firmly in place and there are unseen forces at play. So one day, unsurprisingly, I came down to find this:

You can also see inside that it wasn’t the cleanest operation.

Rubbing salt into the wound, the waterwheel drive shaft decided to break again around this time as well, with the waterwheel itself running aground somewhere downstream. You can imagine how exasperated I was with the whole project. A “working” prototype… which kept breaking.
So, after nursing my bruised pride, I resolved to teach myself to weld. I planned to construct a thicker drive shaft with a heavy-duty welded brace for the waterwheel face/hub, and a welded support frame for the whole enchilada.
A quality welder was going to be a serious purchase, but screw it, I thought – there were other welding jobs that needed doing (barn door latch, etc.), and a little MIG welder is pretty necessary on a farm – especially when you’re relying on hay-making equipment like a mower, tedder, and baler. Not to mention the trailers, quad bike, and horse float which will also indubitably need quick fixes here and there during their lifetimes. It’s also another step towards self-sufficiency, being able to repair and create with steel.
So we treated the purchase as an investment. And seeing as how the proof of concept for getting water to the farm was guaranteed, I might as well do it right and build a unit that wouldn’t break!
I started out by cutting the steel pieces for the hub brace.

Like magic, a super strong hub brace suddenly appeared. I was surprised how easy welding was. I expected it to be painstakingly difficult. It’s actually super fun and pretty straight-forward (with MIG welding, at least), with instant practical results.

After stripping back the mill scale and cleaning up the spatter from my welds, I slid the bearing blocks onto the drive shaft and brace. Then I sprayed it with steel primer.

A few slaps of black to stop it from rusting.

Same process for the frame.



Then once everything was dry, I test fitted it all. Looked a bit too sleek to be going outdoors!

I also welded together an “arm” for the end of the drive shaft opposite the waterwheel, to replace the pulley I had been misusing as a crank arm.

I fitted this piece over the end of the new drive shaft (before painting it), and drilled holes for engineering bolts. Ah, the miraculous drill press – another invaluable tool for DIY.


I also used this opportunity to harden the mild steel around the bolt hole on the piston head, because I had noticed there was a bit of wear from the hardened engineering bolt which it continuously pivoted on. I did this by using a blowtorch to heat the steel up to a nice cherry red, then quenched it. You can see where the colours appear on the steel is the area that I hardened.

It was time to install the new system down at the river, which took a bit of fiddling to bolt it in place on the wooden supports, because the dimensions had changed slightly from the previous build. As you can see, by this stage, it had rained and the river was back up to a decent level.

The trickiest part is always putting on the heavy and awkward waterwheel, which I’ve needlessly had to do about five times to date. Sometimes alone. Ugh.


I replaced the broken PVC pipe acting as piston cylinder, aligned everything, and let her rip. Here it is after it had been operating with this setup for a while, where you can see the piston shaft’s vertical movement has pushed on the lip of the PVC, cracking and warping it, causing a minor leak:
There was a lot of pressure in the pipe, causing the whole pump to shift backwards as well, so I rigged up a quick brace from some brackets I had lying around.

Didn’t work as well as I’d hoped, but it held together for the time being. I think the system successfully operated like this for another few months or so, before the PVC failed again, despite me letting the pump rotate freely to allow for tolerance.
Another major problem with the pump I’d made was that the o-rings were wearing out way too fast, even despite lubrication and being perfectly fitted and sized on the machined piston head. They would need replacing every couple of months, when leaks would begin to spring.
I’d had enough of the inefficiencies of my homemade linear PVC piston pump, so I went digging to find the old colonial farm pump I had originally tried to incorporate into the system. Back to basics, I suppose?

I didn’t have high hopes that I could get it working – and I had no idea if those kind of pumps required a high RPM. I assumed they did. (Spoiler alert, they don’t. Had I known this, or bothered to find out instead of needlessly experimenting, I would have saved myself a lot of time and money. Do your research, folks!)
The old pump was a little worse for wear, broken in places, rusty, grimy, gunked up, and required some replacement parts. Luckily, pumps like this (Davies B1 Series Piston-Pump) are fairly common in NZ, so I was able to source the replacement parts online. The fact that these are still sold new (at a whopping NZ$2,500+) and serviced widely to this day, is a testament to their time-tested robustness and efficacy.
I started by stripping apart the wet end, where the old gaskets had perished and fused to the steel.

The washers and springs were in surprisingly good condition.

Don’t you love messy mid-project bench top shots? I do.

I took an inventory of all the pieces I took apart, and made sure to record where they went, and in which orientation (photographing as you go helps). I removed the steel stand and the platform for the motor, since neither would be needed, and only added unnecessary weight to the already heavy beast.

I took a wire brush to its rusty parts, and de-greased the whole unit with some methylated spirits (denatured alcohol). I also washed the stinky putrefied oil out of the crank end by flushing it with diesel a couple of times, then methylated spirits.

Some smaller components were heavily rusted, so I soaked them overnight in vinegar and citric acid (don’t forget to neutralise the process with baking soda or similar, or the acid will continue to corrode the steel).

They shined up pretty nice.

New replacement gaskets, hooray!

New leather “buckets”, huzzah! These act as the plunger or piston head inside the pump. Being leather, they conform perfectly when wet.

Don’t you love tidy mid-project bench top shots? I do.

I fitted the appropriate pipe fittings to the intake and output, slid a small aluminium pulley onto the crank shaft, and screwed the pump to a wooden base.

Now that the pump was all ready to go, it was time to play with pulleys. I had a very rudimentary understanding of pulleys and how their ratios worked, but I was soon to learn a whole lot through tedious experimentation. Pulleys are expensive, so I wanted to use the assortment I salvaged from the farm. There were a variety of sizes, but annoyingly their bore diameters were all different.

I thought (incorrectly) that I needed to get the RPM quite high to operate the pump, so I did some calculations and devised a ratio that might work.

I reckoned that the resistance of the pump would slow down the waterwheel, so I aimed for a higher RPM than the pump was rated for (200 RPM) to compensate. I wasn’t necessarily wrong, but this was a rookie mistake.
Back at the engineers, I had them machine an axle that would fit a couple of my pulleys and some spare bearing blocks I had from the previous waterwheel drive shaft.
I then welded together a support frame for my ridiculous arrangement, which at the time I thought was awesome.

I also welded together another pulley shaft to go on the drive shaft of the waterwheel.

After a month of fiddling, I had everything I needed, finally ready to build waterwheel 3.0!

Back down at the river, I eagerly installed my vision.

It spun pretty fast! Albeit wobbly. My optimism was increasing.
The V-belt tensioner I devised consisted of a chain and some cabling that could be adjusted on a steel stay.

I built some temporary belt tensioners which I bracketed on to the frame to prevent the V-belts from slipping. The pump was installed, then it was ready for a test!

Aaand… fail.
Didn’t work. I mean, it did, but it slowed the waterwheel right down. Why? Well, it turns out that a pulley is a lever. Yep, that’s right, a lever. Without going into a technical rant (which would likely only further expose my incompetence in these matters), basically what I had designed wouldn’t work because I was trying to get a high RPM from something that didn’t actually have that much power.
So even though I could get the system spinning really fast by itself, as soon as I added resistance to it from the pump pressurising water, all of that leverage brought the waterwheel near to a halt.
I stood back, looked at the over-engineered monstrosity I had constructed, and then had an epiphany. I quickly disassembled the mess and produced this elegance:

Yep. Worked a charm. Why not try that to start with? Idiot.
The ratio of pulleys was quite extreme, and I was starting to understand how they interacted. So I experimented by changing out the large pulley for the medium one, which worked a bit better. It seemed there was a “sweet spot” for the pulley ratios. I knew this one wouldn’t work when the river’s water level would inevitably recede over the drier months, but for now, it was working. And this time it was simple and elegant.

It was especially effective when the river flooded. It was good to see that the new drive shaft brace and frame I welded stood up to a powerful torrent.
As expected, when the river started flowing less, the pulley ratio didn’t cut it. So it was back to the drawing board.
I even built a flexible joint to try direct drive, now that the pump proved to work at a low RPM. A direct drive meant there was no leverage slowing the waterwheel via pulleys, only the resistance of the pump itself.



It did pump to the tanks on the hill, but alas, ’twas not fast enough. More experimentation with pulleys was needed to find that sweet spot.

It was surprising to discover that very small diameter pulleys (as pictured on the pump above) add considerable resistance to the setup due to the V-belt having to wrap around a smaller circumference. There was a lot of energy being wasted on pinching the V-belt around the tiny pulley. I realised I needed to use the larger diameter pulley with a medium sized one for the pump end, so the V-belt’s limited flexibility wasn’t wasting energy.
But we didn’t have the right size or fit on hand. We tried this setup, which confirmed it was just a matter of the right ratio, but the pulleys were just too similar in size:

We tried various combinations of what we had on hand, and it did pump to the tanks quicker than the direct drive, but still not as fast as we needed. We did some calculations of how much each cow and sheep on our farm would need per day in the peak demand of the summer heat. For fifty cows and eighty sheep, the waterwheel needed to be pumping up two litres per minute, for a total of at least two-thousand-five-hundred litres per day.
So I sacrificed one of the smaller pulleys which didn’t have the correct bore size, by welding a bracket of the correct diameter over its bore hole, which I could tighten against the crank shaft of the pump.

Once it was on, it was pumping up the most we had seen. The river was low, so we knew we were on the right track. If only there was a way to make the waterwheel itself turn faster…
With Char’s brain power added to the mix, we spent a weekend experimenting with making wider fins for the waterwheel, which produced instant results.
First we flattened the tin fins (150 x 150 mm) and added wood to either side, widening them to a total of 300 mm.

Then we chipped away at the weir (natural channel in the bedrock) with a wrecking bar to accommodate the larger fins.

We even built a little dam to divert some of the river’s velocity towards the weir, which helped immensely during the drought.

See that devilish grin? That’s my renewed optimism shining through.
It was quite helpful to record the fins in slow-motion, to see how the water interacted with the them hydro-dynamically, and how eddies were being created in places. We noticed the flatter fins produced less turbulence than the previous bent tin fins. Fascinating.
With each small improvement to the overall system, we saw incremental increases to the output at the stock tanks on the hill, which we re-measured whenever we made a minor change. With a “good enough” ratio of pulleys, widened fins, a tailored weir, and a little rock dam, we saw a three-fold increase in output from a measly eight-hundred litres per day from the direct drive to over two-thousand-six-hundred.
That brought us to our daily demand! YUS!

But the output was pretty damn close to our demand. Ideally we’d produce a decent surplus to cover our asses if future droughts were worse (likely).
There was one last thing we could do to push us over that threshold. The makeshift pulley we had fashioned to the pump was producing some inefficiency. It was wobbling due to my quick chop job, and it wasn’t as small as we would have liked. So we forked out fifty bucks and bought the perfect pulley, crossing our fingers that that’s all it would take to iron out the kinks…
And guess what?
That simple amendment took us from ~2,600 litres per day right up to ~3,300!

That truly is a picture worth a thousand words. It gives me the fizz.
Of course now we want to see how much we can squeeze out of it by attaching a dynamo generator to charge batteries… Oh the potential!

We’re thrilled to have such a quaint (and quiet!) solution to our farm’s water supply. Not only does it save us time and money, it has a character that makes us smile. It’s an attraction. Something a bit quirky. It’s a delight to amble down to the riverside just to sit in the dappled light of the forest and be hypnotised by a little waterwheel’s endless dance to the tune of a babbling brook.
Here’s the numbers for the more analytical among you:
- Our peak summer demand for 50 cattle and 80 sheep is 2,500 L per day. Our petrol pump outputs 576 L per hour. So we had to run it for 4.5 hours per day to reach our demand. 4.5 hours costs approximately NZ$10.00 per day. (Since our petrol pump is small and each fill lasts an hour, that’s also 4-5 trips to the back of the farm, jumping fences, hiking down to the river, refilling the pump, etc., every day during peak demand.)
- The incremental improvements to the waterwheel pump were:
- Direct drive: 864 L/pd
- 560mm & 360mm pulleys: 1,584 L/pd
- 560mm & 65mm pulleys: 1,872 L/pd
- 560mm & 260mm pulleys + dam + half of fins wider: 2,300 L/pd
- 560mm & 260mm pulleys + dam + all of fins wider: 2,664 L/pd
- 560mm & 200mm pulley + dam + all of fins wider: 3,312 L/pd
- Surplus generated during peak summer demand: +812 L/pd
- Disregarding the “educational costs” of the waterwheel’s previous iterations, the estimated cost for the current waterwheel setup (excluding the free pump and large pulley) is NZ$500.00.
- ~$100.00 steel
- ~$100.00 wood/concrete
- ~$40.00 hardware fixings
- ~$60.00 pipe fittings
- ~$100.00 bearings
- ~$100.00 pulley/V-belt
- That means that the waterwheel only needs to pump for 50 peak demand days to pay itself off, since we’d otherwise be paying for petrol. That’s a no-brainer. It would only have to run for a year to pay itself off if you include the price of a new piston-pump worth NZ$2,500.00 (and that kind of pump is overkill).