Low flow matters.

I measured available flow again this morning.  There hasn't been any significant rain, so I wasn't expecting it to be any better, and it wasn't, - only 0.9 lps.

0.9 lps is 30% of the design flow for my scheme (3 lps) and from the graph in my last post, it is clear that a pelton is still pretty efficient at this part flow.  So why can't I be generating with the 0.9 lps available ?  Why is it I'm waiting until flow increases to 1.2 lps ?

The answer lies in what happens to the output voltage of the SmartDrive at low flows.

There are two determinants* of the output voltage of any permanent magnet alternator (pma): rotational speed (rpm) and load. Thus:

• a decrease in rpm will lower the voltage
• a decrease in load will raise the voltage

Now the first of these, rotational speed, we can discount as causing voltage change: the rpm at which the pelton and alternator turn are more or less constant and determined by the characteristics of the hydro half of the installation.  Ultimately it is net head which fixes the rotational speed and because net head won't change, rpm won't either. For my set up, it is 1,200 rpm.

So any change in voltage has to result from the second determinant: a change in load, and the only load on the Smart Drive is the inverter.

Recognising that the inverter and pma interact with each other is a key understanding. The inverter, far from being a fixed load of so many ohms, is instead a smart device which can change the load it puts on the system. This is seen in two ways:

1.  a steady 'hunting' of voltage as the MPP tracking function of the inverter continually seeks the optimum voltage and current combination.  This can be seen in the following plot where a SmartDrive with nominal output voltage of 220v dc is seen to hunt up as far as 285 v and down as far as 200 v over the period of time the record was made.

2.  a step change in voltage when nozzles are changed to generate either more or less power according to available flow.  Smaller nozzles generate less power and cause the system voltage to rise. This occurs because the inverter, working in tandem with the pma, presents itself to the pma as a reduced load.  

Thus in my scheme, when power out to the grid is high: 713 W, system voltage is low: 295 ~ 315 v dc (the variability being due to MPP hunting as above), but when ac power out is down: 240 W, system voltage rises to 360 ~ 379 v dc.

The effects of these higher operating voltages are several:

• higher transmission voltage so less line loss - GOOD
• higher dc input voltage to inverter reduces its efficiency - BAD
• dc input voltage rises above inverter MPPT voltage range (100 v ~ 320 v) - BAD
• Powerspout's V-clamp starts to dump some of the generated power - VERY BAD

So, with the final point, we have at last arrived at the main reason why 1.2 lps is the lowest flow I seem to be able to operate at, - any less and the system voltage rises to be too high, too much of the time: the V clamp spends so much time dumping power that only a small amount (150 W at 0.83 lps) ever gets into the grid through the inverter.

Here is a record of the power output into the grid at two flow levels: 1 lps (237 W) and 0.83 lps (150 W), showing some dumping at the higher level, evidenced by the occasional, random, downward spikes, and at the lower level, the pattern created by more frequent dumping:

Perhaps I should consider buying a different Smart Drive core for use at low flow times of the year: a core wound so that its output voltage stays around 300 v dc when power generated is in the range 100 ~ 260 W.  

They're available from EcoInnovation at $US 199 plus freight.

And that's what I like about this game, - always something new to consider !  Always a new reason to spend !

*There are others, eg magnet field strength, width of air gap between stator coils and rotating magnets, number of poles and whether connected delta or star, but these are not variables in a Smart Drive pma which is installed and functioning.

One jet or two ?

In their literature, EcoInnovation recommend that Powerspout peltons are not operated on one jet if the power output from the Smart Drive exceeds 400 W.  This is presumably because of mechanical stresses on the root of each runner cup.

For a GE 400, (or the later Powerspouts which are directly grid connected using an Enasolar inverter, without the V-clamp board which is found in the GE 400), 400 W DC output from the Smart Drive equates to about 330 - 350 W AC from the inverter to the grid. The actual figure will be dependent on transmission losses and inverter efficiency at this power level.

The plot below is taken from the literature of a long established UK manufacturer of water turbines based in the Lake District and shows how, for their turbines, there is an efficiency benefit to be gained from delivering the flow via two jets rather than one when the flow is toward the low end of design flow.

Assuming what is true for their twin jet peltons is also true for a Powerspout  and applying the information from this plot to my installation, it would mean that at 40% of full flow (1.2 lps / 3 lps) I would gain 3 % in efficiency.  In power terms this equates to seeing 250 W rather than 243 W into the grid.

So the message would seem to be: at those levels of flow where EcoInnovation say it is OK to run on one jet, it is actually more productive to still operate with two.
Correction added 28 May 2015: this is an erroneous conclusion. Please refer to later addendum "One jet or two - the bigger picture"

I have to admit though, - I haven't tested this experimentally.  It would be very difficult to cut 3 nozzles sufficiently accurately to ensure that the sum of the flow of two of them was exactly equal to the flow of the third.

In the next post: since the above plot shows that a pelton remains reasonably efficient down to a part flow of 20%, why can't I operate mine below a part flow of 40% ?

A conundrum solved, - possibly.

Throughout the first year of operation, there had been something niggling me and needing an explanation:  I had noticed there was a difference between the power generated by the top nozzle on its own compared with when the same nozzle was in the bottom position.  In the top position, the power from the inverter into the grid was about 20% less.

The factors influencing the efficiency of a pelton in converting energy in its water jet into rotational energy in its shaft, constitute a complicated science, but one of the key factors is getting the jet to hit the buckets of the wheel (also called a runner) in exactly the right spot.  

What I discovered on dismantling the Powerspout for its summer service was that the positioning of the centre of the shaft was not at the midpoint between the two tangents which the jets form with the 'pitch circle diameter' (pcd) of the pelton runner.

To be sure of the line taken by each jet to make the above measurements, each jet was given two nozzles, one inside, the other outside the casing, each having small diameter holes. In this way parallax error was minimized when sighting through them.

I have come to the conclusion that this asymmetry accounts for the power discrepancy I had seen, yet just as soon as reaching it, I doubt it, simply because the folk at EcoInnovation who designed the turbine know their stuff and are unlikely to have allowed such an error of geometry.  In part it could be explained by there being some play in the nozzle holders where they pass through the casing: by positioning the holders at the extremes of their play, the difference in the dimensions as given above can be made smaller, though never eliminated.

As can be seen above, it is not obvious that the dimensional difference is visible from the splash pattern of the two jets.  Where it does become obvious is when changing nozzles: the clearance between nozzle and runner for the upper is less than for the lower, making it more fiddly to do.

It would have been a simple matter, before having the new stainless bulkhead cut (see last post), to adjust the CAD drawing so the shaft and bearing housing were better centred, - but I hadn't solved the conundrum before having it cut.  

So it is going to have to remain as one of those efficiency losses which are an inevitable part of all machines.  To minimise the effect of it, I make sure that I always have the bigger of the two nozzles on the bottom where it will produce better output.  

In the next post, we'll look at the merits of one nozzle vs two nozzle operation.

Summer maintenance

The hoped for rain has not materialised !  Flow from the spring is stuck at 0.85 lps.  So with the start of generation for this 'water year' not looking imminent, I'm going to back-track to  August to say something about the yearly maintenance done then.

It was interesting to see what effects one year of operation had had. Impressively there was no wear detectable on any of the nozzles or runner buckets and I put that down to silt mostly dropping out in the header tank, from which about half a cubic metre had to be bucketed out.

I had been concerned that silt might also collect at the bottoms of the upstand pipes leading to each Powerspout nozzle.  To clear the pipes through, the following arrangement was used but from the colour of the discharge, it didn't look as if much silt had collected.

The only serious concern evident on dismantling the turbine was galvanic corrosion going on between the aluminium bulkhead and the stainless steel dump load element, and also where the stainless self tapping screws secure the plastic shell to the bulkhead:

Whilst none of this corrosion would likely have caused failure in the near future, certainly it would have needed attention eventually and I decided to replace the bulkhead with a stainless version.

Michael Lawley in New Zealand was immensely helpful as ever and sent me the CAD drawings to get the replacement cut.  I had never had any experience of laser cutting before so the accuracy which is possible, - to 0.2 mm, astounded me.  

Here is the finished article after refitting in the casing, using M5, socket-head machine screws rather than the original self tapping screws, and a 1" BSP stainless nut, which has a 'captive' O-ring behind it, to hold the heater element: (the bulkhead cost £60 plus £12 VAT)

To complete the annual overhaul, a new set of bearings was put into the bearing block.  Ever helpful again, the EcoInnovation instructional video on this was first class.

That's enough for this post.  I'll add a few other snippets discovered during maintenance in the next.

"Capacity factor" and "Water Year"

Capacity factor and water year are two useful concepts.

The capacity factor for a scheme is a measure of its productivity over time.  It is calculated from the actual energy generated over a year as a proportion of how much energy could have been generated if the scheme had run at full capacity "24/365".  Like overall efficiency, it is one of those things which would be nice to know for a scheme before it is implemented, - but cannot be with certain accuracy.

No scheme can run 24/365. Most 'run-of-river' schemes have to reduce generation or shut down in the drier months for lack of flow. Then there is down-time for maintenance, breakdowns and grid-out spells.

For a Powerspout installation there is the added factor that nozzles have to be changed manually. As the flow changes with the seasons, so the size of the nozzle has to change.  It is somewhat of a "hands-on" business doing this, with quite a learning curve for reading the signs as to when to make a change, but the assiduousness of the operator in doing it  will improve the capacity factor. 

For my scheme in 2013/14, the total generation was 3,672 kWh.  The peak power output possible in that year was 0.713 kW which makes for a maximum possible energy generation of 6,245.9 kWh. So the capacity factor was 0.59.  My hope is to better that in this next year.

Which leads on to what period of 12 months is best taken as the period of time over which to base such a study: a calendar year ?  a financial year ? an Environment Agency year (which runs 1 April to 31 March) ? - or a "Water year" ?

Since the operation of any hydro site falls into a natural rhythm governed by the seasons and rainfall, it makes sense to use such a natural pattern and call it a "Water year".  It should begin and end in the middle of the driest season.  Here, that runs from August to October, so my water year starts 1st October and ends 30th September.

One advantage of doing things this way is that if there is a particularly severe dry season, and they happen every few years, its effects on the output record will be spread over two water years, rather than giving one apparently very bad year.  It only makes the figures appear to be better !

Down hill all the way.

I've just measured available flow this morning and it's up to 0.85 lps, so getting closer to the minimum of 1.2.  

Whilst the turbine remains inop therefore, let me continue the discussion about efficiency:

Having determined by experiment the overall efficiency to be 0.5 (see last post), one can speculate as to what the component efficiencies contributing to this overall figure might be.  Bear in mind though: these are either estimates, culled from the literature as being 'typical figures', or calculated efficiencies specific to my installation.  Their only justification is that collectively, when multiplied together, they give the overall efficiency established by experimental means: 0.5.

• penstock efficiency @ 3 lps                      0.95
• manifold efficiency                                    0.98
• nozzle efficiency                                        0.98
• Pelton turbine efficiency                           0.77
• turbine / alternator drive efficiency         1.00
• SmartDrive PMA alternator efficiency    0.80
• transmission line efficiency                     0.98
• SMA 1200 inverter efficiency                  0.90

What happens to these different efficiencies at flows below the design flow, which for my scheme is 3 lps ?

Some will improve, such as the hydraulic related components and the transmission efficiency, but the pelton efficiency and the inverter efficiency will suffer, most especially the latter.  Let us see why:

Below is a copy of SMA's efficiency plot for the Sunny Boy 1200 and on it I have annotated real data taken from my scheme's first year of operation.  What can be seen is that when PAC[W] is low, the efficiency drops off markedly.  In fact at all points, the inverter comes nowhere close to operating at its best efficiency because the input voltage, which is between 300 and 379 V DC depending on power generated, is not the optimum voltage which the inverter likes to see.  

SMA SunnyBoy efficiency curve:

So the message is: at those times of the year when there is low flow / low power generation, the efficiency of the scheme slides down hill rather fast.  Just when you want to get as much power from your diminished flow as you can, you end up getting proportionately even less.  

To put it another way, for my scheme it takes 18 m³ of water to generate one kilo-watt-hour when the flow is 1.2 lps but only 13 m³ when the flow is 2.7 lps.

Cutting to the chase: overall efficiency is just 50%.

Most people grasp that the power output (P) of a hydro scheme depends on three things: 

• the head of water - the height difference down which water is flowing (Hgross)
• the flow available (Q)
• the overall efficiency of the scheme

... and so it is.  But actually the head component has its effect because it determines the pressure seen by the turbine's nozzles.  To convert head into a pressure, mathematical constants need to be employed to account for the acceleration due to gravity and the density of water.

So the equation for power output, incorporating these constants and adjusted so the units are those of kW, litres/s and metres becomes:

P = 9.8 × Hgross × Q × total efficency

1000

From my records of the operation of this turbine in its first year, I know that when the flow was 2.72 lps, the electrical output to the grid was 0.713 kW.  I know this with some confidence because there was a period of 75 days between January and March when the turbine ran continuously at this flow and the Elster export meter (which reads to 0.1 kWh) displayed the accumulated energy generated over this time. By dividing the kWh figure by 24, the number of hours in a day, the instantaneous power output is given, and it comes with the assurance that it is a mean figure arrived at over a lengthy period of time.

I have measured the gross head (Hgross) for my scheme as 53.6 m using the Budenberg gauge pictured in the last post (though I have to admit its calibration has not been checked and being an eBay purchase, it could be out).

Entering into the above equation the knowns for power (0.713), flow (2.72) and gross head, and solving for total efficiency, it turns out to be 0.499, or near enough 50 %.

The phrase "water to wire" efficiency is often, and usefully, used to describe this overall efficiency figure for a hydro electric installation.  It is a crucial determinant of the cost / benefit ratio for implementing a scheme, but unfortunately it is almost impossible to know with any certainty before the scheme is up and running.

In the next post, I will look at what factors in my installation contribute to this relatively low figure for "water to wire efficiency", and what happens to its value at flows below 2.72 lps.

Accurate to two decimal places.

What I like about operating my Powerspout is both the unpredictable element, namely the dependance on the vagaries of rainfall, yet the very precise predictability of how much power will be generated for a given flow.

So I have taken to operating my installation rather like a laboratory experiment.  I measure, record and calculate various inputs and outputs which enable me to better understand the quite complicated engineering which underlies such an apparently simple machine.
  
Most especially what this admittedly 'anorak-like' approach shows is how losses, or inefficiencies, in the system drastically reduce the electrical output.  I want to draw attention and explain this further in the next posts on this blog.

To make any kind of scientific investigation, I have had to decide just how accurate I intend to be, and an accuracy of two decimal places is the tolerance I aim for in most readings.  This should mean the conclusions drawn are reasonably valid.

To measure available flow, I divert water from going into the header tank and measure with a stop-watch the time it takes to fill a coal scuttle (11.36 litres).

Today, it was 20 seconds, the same as yesterday, making the flow 0.57 lps.  Still not enough to run the turbine continuously without draining down the header tank.  The premise on which my scheme works is that outflow from the tank cannot exceed inflow if the turbine is to operate continuously. The corollary of this is that the tank must always be at least slightly overflowing.

The out flow from the tank to the turbine is determined by the size of the two nozzles jetting water onto the pelton wheel.  The out flow can't readily be measured but it can be calculated. The formula is: 

Q = CD × Anoz × √[2g × Hn] 

where Q is flow, CD the discharge coefficient for the nozzle, Anoz the cross sectional area of the nozzle hole, g is acceleration due to gravity and Hn  is the net head of the scheme.  

So three unknowns: Hn, Anoz and CD have to be discovered.  Here's how I have measured each:

1. Net Head

2. Nozzle area

3. Discharge coefficient

Powerspouts are provided with well designed, 40° tapered nozzles having a discharge coefficient (CD) of 0.89-0.91.  I established this value one afternoon by using the same equation above, but this time in-putting the flow as a known. I had somewhat laboriously measured it by measuring the turbine discharge flow. (note added 6 Feb 2018: the Cd was later measured to be 0.85 not the figure given here; see this diary entry).

Solving the equation for the two nozzles fitted in the turbine at the moment, I know that the outflow will be the sum of 0.83 and 0.30 lps, ie 1.13 lps.

Since this is twice what the inflow is, it is clear continuous operation of the turbine is still not possible.

Patience is needed, but the good news is that the weather forecast is for a warm, wet week ahead.

Time to start a blog ...

The summer is over, leaves are falling from the trees and rain is also steadily falling from a leaden sky, - all factors I take note of as together they mean ground water will be building up. Soon there will be sufficient gushing from the spring which is the source powering my turbine.  
This start to another "water year", the second year of operation of my turbine, seems an opportune time to start a blog about it.
I need a minimum of 1.2 litres per second (lps) and this morning it was just 0.57 lps.  So some way to go yet. But with the steady drizzle and rain we're having at the moment, the flow should be there in a week.
Yet I have come to learn in its first year of operation that you cannot predict anything. You are entirely in the lap of the weather and being so creates a humble appreciation of the natural world.
Hope and expectation spring eternal nevertheless... the hope being to better the first year's output figures: