Sherborne Mill is now generating!

Harry Clarke
Photos by Keith Andrews

As a recap, our aim here on the historic site of Sherborne Mill has been to establish a power-generating waterwheel and incorporate it within a building that is sympathetic to the setting.

We have taken the construction slowly.

With accreditation secured we then halted the wheel and lowered the mill head during the summer just gone to allow the construction of the building to take place around it in safety. Although we had run the wheel for three months during 2019 the top ‘soldier’ brick course of the repaired leat walls had not been in place. This meant we had not been able to bring the water level in the mill head right up to full working height. This reduced flow meant that the maximum power we had seen in 2019 from the wheel was about 2.2kW. Clearly there was more to come.

Syrupy stuff

Over the New Year I therefore re-installed the electric synchronisation box (a HGP2 system supplied by Sustainable Control Systems) which had been kept away from the dusty site. I also serviced the wheel by re-greasing the taper roller bearings seated within the two plummer blocks in which the axle turns. Finally, I changed the oil in the double epicyclic speed multiplier gearbox – 3 litres of fully synthetic EP220 with much the consistency of golden syrup.

Exciting times

Obviously the wheel starts to turn and with it the generator.  There is however quite a lag in the process.  In fact it takes about 45 seconds between adjusting the logs and the wheel settling down to a revised speed based on the adjusted flow.  Once the wheel starts to turn at about 5rpm the theory is that the self-excited generator starts producing a voltage.  (As some will know, self-exciting generators rely on the iron cores on which the field windings are wound retaining some magnetism from the last time of operation). When starting from rest, and so electrically dead, this residual magnetism nonetheless primes the creation of an initial voltage which then drives further electromagnetism via the field windings, and off it goes.

Anyway, we duly did this but got nothing. After much head-scratching, reading of manuals, and phone calls it became clear that we had to ‘re-flash’ the generator. This is required with any self-exciting generator that has been stationary too long and from which the residual magnetism has ebbed away. The process is pretty much as it sounds. Mains voltage is applied directly to the field windings, which, being made of thick copper, offer very little resistance. The instant current draw is huge making the eponymous flash and blowing apart any fuses unwise enough to be in the way. Quietly, without mentioning it to the family and armed with insulated pliers, welding gloves, and a mask, I set to. The results were threefold:

• Devika came rushing out wondering what had happened, all the household lights having dimmed;
• the 40amp generator trip popped;
• it worked!

Slightly tediously however the next day my laptop on which I am writing this, and which had been charging at the time in the house, wouldn’t boot up. It subsequently required a £500 new motherboard. Yet another expensive mill owning lesson learnt!

Synchronised spinning

Anyway, the next day, with the generator freshly re-magnetised we again slowly fed water to the wheel. Duly, at around 5rpm, the control box first began to register a rising voltage. The working speed of the wheel is a constant 8rpm – at which point the geared generator is rotating at exactly 1500rpm and producing a 50Hz sine wave. This is of course exactly the same shape and frequency as that produced by the National Grid.

By increasing the wheel speed slowly the voltage rises further. Once it gets to 160V (the wheel speed being about 7.6rpm at this point) the control box display also starts, showing the frequency of the output sine wave. Typically it will begin at around 47.0Hz. The trick, at this point, is to cautiously and incrementally crack open the stop logs further so that the speed creeps up very slowly until the wheel is running at a speed which means the generator is producing as near as dammit 50Hz. At this point the clever electronics and fairly hefty capacitors in the control box somehow manage to buffer the two almost synchronous but inevitably out of phase sine waves (the one from our National Grid connection and the output from the wheel) so that they fall into both phase and frequency then with a big ‘kerchunk’ of a monster relay and shudder from the wheel, the two are connected.

This was duly achieved.  From this point, all need to be gentle with the stop logs evaporates.  We simply remove them all, giving the water in the mill head an unobstructed path down the leat to the wheel.  More water does NOT however make it go any faster. The National Grid has effectively got a grip on the wheel at this point and will not allow the generator, to which it is geared, to rotate at anything other than the speed which precisely supports a 50Hz sine wave.  More water does however create more torque on the axle, more current and hence more power.

Satisfied with this, with the mill head up to working height for the first time ever and a steady 2.68kW showing on the output meter, we left the whole lot running nicely.

The spillway in the mill pond regulates the water flow with a clever sloping cill. 
The excess water runs off into the bypass channel, which can be seen on the right of the picture at the start of the article.

Ripping yarns

That night I awoke. Despite the gentle noise of the rain it was clear that something else louder was going wrong. Jumping up and to the window – our bedroom faces the wheel across a courtyard – I could see the wheel spinning fairly wildly and obviously without any restraining load. Having rushed about semi-naked outside in the dark for a while, pulling stop-logs from the control sluice and putting others back in the leat, I went into the generator room. Although the wheel is coupled to the gearbox, and hence the generator, it is not done rigidly. Between the wheel and the gearbox is a ‘Fennerflex 140’ coupling which looks much like a wheelbarrow tyre.

One bead is gripped by a flange on the waterwheel shaft and the other by a flange on the gearbox input shaft. As a result, drive and torque is transmitted through the fabric-reinforced rubber of what would be the circumferential tread of the ‘tyre’.

Stormy shakedown

Assuming it was bad luck, I ordered another rubber replacement.  Later that day though, dusting off my torque tables, it became clear that our wheel, under load, was more than capable of exceeding 3kNm and in fact would probably operate around a constant 4.7kNm when the buckets were fully loaded with water. 

Having ordered a new F140 tyre I fitted it, as a stopgap and at reduced power, only to find that it lasted less than 72 hours.  The full solution – an F180 coupling (which is substantially bigger and of course another expensive mill-owning lesson) was inevitable.  This was fitted on 20 January and whilst the thick rubber is looking a little strained when under full load, in the time since it is bearing up well.  Maximum registered load on the control box with this configuration has been 3.31kW which, as you would hope, is pretty much the same as the maximum rated capacity of 3.41kW which shows that all is well.  With Storm Ciara having deluged our catchment and Dennis now on the way it’s certainly getting a good shakedown.

The Generation Game

Although it might seem a bit wasteful not to have had the wheel running during the summer of 2019, I am quite pleased we didn’t.  Despite SSE, our energy supplier, assuring me that our Economy7 meter would be fine to cope with measuring the generated power – it wasn’t.  It didn’t just fail to shut down the registers when we were feeding power back to the grid;  worse still, it actually counted the exported power twice!  Once as it went through the day-time register and again as it somehow flowed backwards through the night-time register.  We are currently in dispute for about £600 (expensive lesson number 3) but we may get somewhere I hope.

This year though I have been keeping a close note of the wheel’s generating performance and it’s looking good.  Taking the period in which the wheel was working continuously from 20 January when the F180 coupling was fitted to today (13 February) we have, on average, generated 58.8kWh every day and had to pay for only 1.8kWh (which is effectively only those odd times when our generation level is exceeded by the demands of the house).  I suspect we are exporting about half of our generated power which, whilst good for society, seems a bit of a needless loss whilst we are still paying handsomely for petrol.  Watch this space!