Enecsys Monitor kWh vs. Generation Meter kWh

[ovonrein]

I had thought of such beasts [MCBs] as more or less passive in terms of their energy consumption?

Yes, completely agreed.  My suggestion was more with regards to some faulty wiring (loose cable etc).  Of course, it is still possible that a cable inside one of the connector blocks hasn't been screwed down properly.  Whatever it is, and where it is, it should betray itself through (significant) heat loss.

[HPSauce]

Hi Jez54. I have a similar system to yours except with 13 Sanyo panels, installed on November 15th.
There is full tree shading at this time of year.

Figures from Enecsys monitor are 38.4kWh total and are much more optimistic compared with the Elster generation meter 28.4kWh reading.
So in my case, the Enecsys monitor is 35 percent optimistic. I contacted our (very good) installers and the response was as follows
"Regarding the Enecsys monitoring software, this is a performance indicator and not a calibrated OFGEM meter, so it is not unusual to have a discrepancy either way. I should think that once we've had the benefit of some better weather then the gap percentage difference should narrow."

My AC run is fairly short, probably 15metres so AC losses cannot explain the discrepancy. I don't believe that a poor connection can explain the differences but we'll see if something starts smoking when the summer comes.
HPSauce

[series530]

As the data on my website shows, I have a couple of rather obstructive chimneys obscuring three of the fourteen panels on our system at this time of year. Even so, there is less than a 2 percent difference between the Enecsys data and the AC100. I find the statement offered by the installer rather astonishing to be honest. I cannot believe that a micro inverter company would knowingly offer a measuring system that is so badly calibrated. The caveat offered on their beta monitor is something of a catch all and I very much doubt that it had a figure of 35% in mind. I also fail to understand why, on days like these, when a system is producing 20% of what it may produce on a summers day, it will suddenly become that much more accurate just because the sun gets a little hotter and the days a little longer.

[HPSauce]

I guess there can be three explanations why my Enecsys data differs so much from the generation meter.
1. Inaccuracies in the Enecsys system.
2. Wiring fault
3. Generation meter inaccurate.

Regarding option 1., Enecsys offer no specification for the accuracy of the monitoring data. If the power is sensed by using the DC voltage drop of a length of copper track (cheapest and simplest way), this will vary by over 20 percent from 10 to 70 degrees C. The system is probably optimised for accuracy at full power. Enecsys also seem to have moved their production facility recently which may have lead to a change in board supplier.
None of my panels has direct sunshine at this time of year so they are only working at 10-20 watts even on a bright day.

Option 2., I regard as highly unlikely since the electrician won an installer of the year award from his previous company and seemed to be highly competent.
My installers were also one of the early adopters of micro-inverters and have plenty of experience with them.

Option 3. is possible but highly unlikely.

I am not too concerned about the differences but will see if the metering will agree more closely as we move toward the summer months.
As I see it, the important thing is to be able to monitor each panel and to see the effect of  shading as in your excellent shading analysis post on the Electricians Forum.

It would be interesting to hear from other micro-inverter users about their findings.

[Iain]

Hi HPSauce

Option 3. is possible but highly unlikely

I had problems with my import export meter so you can't assume the above. Not likely but a possibility.

http://www.navitron.org.uk/forum/index.php/topic,14295.0.html

Iain

[HPSauce]

Hi Iain,

An interesting saga. Maybe I need to invest in an electricity monitor. I only have a plugin one at present.

[series530]

Indeed, a very interesting read. Thanks also for the positive endorsement for my shading analysis. It's available on this forum as well.

On the subject of current monitors:

Just for information, I'm finding that the EnviR/Optismart isn't doing a bad job of reading the power produced. I have a little under a month worth of data so far and the EnviR claims that we have generated about 60KwH during that period. The Enecsys website claims we have generated 64.9KwH. The EnviR seems to under read when the sun is low - the panels seem to sit at a quiescent system power of 196W with them all just turned on. The EnviR reads about 168W under these conditions. Given that there has been quite a lot of low powered sun during December it may account for much of the difference - some of the other is probably loss between the inverters and the PV meter.
If there was a static reading error (ie. an offset) of 30W between each unit, the errors would appear more significant under lower power generation conditions. If that error is constant as we move into periods of higher irradiance, the error term will have far less effect.

No matter how I look at, the very fact that the EnviR registers zero power generation at night against the 268W that Wattson claimed just goes to show that the Wattson current clamp method is just an very rough indicator and little more.

BTW: So far, with our system, Enecsys claims 142.5KwH against the meter reading of  135.8. Just a shame I didn't have the EnviR system on day one as well ! 



 

[HPSauce]

Hmmm, just looked at the Elster generation meter now (3:20pm and fairly bright conditions here) and the red light is on. I think this means no electricity being generated. Yet the Enecsys web monitor says 162W is being produced. I switched off everything in the house and the main electric meter was not showing signs of any electricity flow.
Not sure what this means but it seems to imply that the Elster meter will not allow current to flow below a certain, rather high, threshold. I have  sent them an email to query this.

[jez54]

Hi, all, and thanks for your ongoing suggestions / ruminations about my problem.

I have continued to monitor my installation (every evening manually copying from the Enecsys monitor to a spreadsheet the generation of each panel during each hour of that day, and comparing the day's total with the day's total shown on my Elster A100C meter).  My situation is unchanged since my last post:  the total generation as per the Elster A100C meter is still only 76% of the total generation as per Enecsys Monitor. And still no replies from my Installer or Enecsys or Elster.

Like series530 I cannot believe that a new high-tech company like Enecsys, which was a spin-off from Cambridge University technology labs, would allow such a dramatic variation in Monitor results to jeopardise their marketing efforts?

First, on the basis that the cause might be at the Enecsys end of the wire, I postulated (as ovonrein and others have suggested) that the Monitor may be looking at the DC side of the inverter, and that the AC side may not "switch on" until some fixed input threshold is reached.  I tried simulating a "threshold" model in a spreadsheet, in 2 ways.  Method A subtracts a certain fixed output "X" from each panel's generation as per Enecsys Monitor, taking any resulting negative outputs as zero.  Method B simply ignores the generation of the panels which do not reach the fixed threshold "X", but takes 100% of the output of panels generating more than "X".  I used Excel's "goal seek" function to find the value of "X" which cancels out to zero the error between the total Enecsys Monitor generation to date and the total Elster meter reading to date.
Method A, threshold "X" =   3.6 Watts/panel ( 3.6 Wh/hour/panel .... not quite the same but the only data I have)
Method B, threshold "X" = 13.0 Watts/panel (13.0 Wh/hour/panel)

I do note from my Enecsys Monitor data that 13 or 14 Wh/hour/panel appears to be the generation of those (Sanyo HIT 250W) panels that are fully shaded from direct sunlight, but fully exposed to unclouded "blue sky" overhead and in the direction of the sun.  It is not possible to determine the lowest output of the panel from the data, since I observe plenty of 1Wh/h/panel readings, which is the lowest value that Enecsys Monitor displays.

But no such explanation could justify the differences between the 2.4% loss that series530 observes, the 24% loss I observe, and the 35% loss that HPSauce observes, IMHO?

So then I assumed instead that the Enecsys Monitor readings are on the AC output side of the microinverter and are accurate, so that the issue is at the Elster meter end of the wire.  I simulated 2 more spreadsheet models.  Method C assumes the Elster meter records zero generation until the output exceeds "Y" Watts, and then records only the excess over "Y".  Method D assumes that the meter records zero when the output is below "Y" Watts but then records 100% of the output when greater than "Y". Again I used "goal seek" to find the most likely value of "Y".
Method C, threshold "Y" =   50 Watts  (0.05 kWh/hour total system generation, corresponding to approximately 0.2 Amps)
Method D, threshold "Y" = 200 Watts  (0.20 kWh/hour, corresponding to approx 0.8 A)

I am much more inclined to believe that the "fault" (or "threshold" issue) is in the Elster meter, or perhaps that the way in which the meter accuracy is specified and measured is misleading?

Elster say their meters come in 2 flavours:  1% and 2% accuracy versions.  One immediately thinks of that as meaning the meter is accurate to 1% or 2% of the generation output being recorded at any moment.  But, thinking back to my schoolboy days, I'm now wondering whether, in fact, like an analogue meter, the accuracy is in fact a % of "full scale deflection", that is 1% or 2% of the maximum meter rating?

The only spec sheets I have been able to download from the Elster website say that the current range of the meter is either (BS spec) 10-100 Amp or 20-100 Amp, or (DIN spec) 5-65 Amp or 10-65 Amp.  But, on my Elster meter the spec plate says that the rating is 0.5-10(100) Amp, whatever that means?

If the so-called meter "accuracy" was in fact respresented by a "starting threshold" of 2% of the (apparent) 10 Amp continuous rating of my Elster meter, that would correspond almost exactly to the 0.2 A threshold found by my Method C spreadsheet model.

Or, if the meter accuracy was 2% of a 100 Amp peak rating, that would be +/- 2 Amp, which would be a very significant 12% proportion of the 16 Amp peak current when my 4kW PV system is generating its maximum output.  And (given the very reduced ouput this month due to low sun and shading issues) such a meter error would be of a similar magnitude to the maximum output of 0.6 kWh/hour = 2.5 Amp @ 240 V which I have generated this month.  So that, the so-called "accuracy" of the actual meter readings could be totally swamping the expected meter reading, making any recorded value potentially wholly random - and reducing my FIT entitlement by up to 24% as a result?

Plenty of food for thought, anyway ......

Other ideas still welcome, because I'm not at all sure I have bottomed this out yet ...... and I'm not prepared to give up 24% of my FIT tariff without a fight .....!!

[series530]

Your post got me thinking:

In meter technology, we do, indeed, normally specify accuracy, in terms of percent of full scale reading together with a fixed offset and some gain error.

Looking at the Elster specs  they seem to refer to something called Rated maximum current (of 100A) and then a term called basic current (ib) which is either 5, 10 or 20A. At first, I assumed that this was a set of current ranges but this doesn't quite make sense because it implies that anything above 20A exceeds the maximum current range.

Anyway, taking this for a moment, it says that, depending upon class of usage, the starting current is either 0.004 or 0.005 of ib and that the accuracy range is ib/20.
I have to confess that my background is in low power electronics and this type of specification is somewhat alien to me. Let's however, look at a worst case: we'll work on a 20A range (16A being the maximum that our systems should be producing): the starting current should be 0.005*20A =100ma. Let's assume that each inverter is producing 1W with, lets say 240V AC output. If we had 16 panels running and ignoring power factor, that's a current of 4ma per panel or 64mA across the whole array. My panels appear to turn on with at least 10 times that as a minimum, so we should be seeing at least 0.5A on a really dull day. That is far more than the starting current requirement of ib/20.

What isn't specified is the linearity and offset errors. I find it hard to believe however, that they could be so bad as to contribute the errors that you are seeing. The kit that we make, for example, has an an accuracy of +/-0.1% and a resolution of 0.0002% or range.

Just some ramblings - perhaps I'll come up with something more after the new year is in.

[Wickham]

For what it's worth, I have Enecsys micro-inverters and I've just checked the total compared with the generation meter again after a dull December when the gateway unit was often showing 17 (ie 16 plus the gateway unit which means they are alive compared with just 1 when the micro-inverters are closed down) whereas the generation meter was often showing a red light at the time ie not generating.

The difference was 1640.4 compared with 1713 for Enecsys ie a 4.4% reduction. This has been slowly increasing from 3.9% a few months ago.

December on its own was 57.1 compared with 63 for Enecsys, a 9.36% reduction for a very dull month.

When the micro-inverters first switch on in the morning, all show 10Wh while the sky is still dull, but when the sky gets dull in the evening they go down to 9Wh before closing down.

It just shows that your high percentage must be a serious issue, but I can't help I'm afraid.

[ovonrein]

December on its own was 57.1 compared with 63 for Enecsys, a 9.36% reduction for a very dull month.

That would be consistent with the Enecsys monitor reading the DC input and the micro-inverter having an efficiency of ca 92% - reasonable.  Interesting thing is, jez54, if it is indeed so that Enecsys report the DC side (only), then it is awfully hard to identify a faulty inverter.  You might have to run this for 16 days, each day with a different inverter disconnected from its junction box (always comparing Enecsys monitor to generation meter).  PITA.

When the micro-inverters first switch on in the morning, all show 10Wh while the sky is still dull, but when the sky gets dull in the evening they go down to 9Wh before closing down.

I think that's normal.  The switch-on threshold may be set higher than the cut-out threshold.  The same on my SMA.

[HPSauce]

What an intriguing puzzle....

Thanks guys for your detailed posts.

A few further observations.
When the Enecsys website was reporting 175Wh production, the Elster meter indicated 1 pulse in 55 secs thus 65Wh.
This would indicate a loss of (175-65)/13 = 8.5Watts per panel.
This seems like a high value for the quiescent power but could be possible. It would contribute a 3.4 percent loss
at maximum power.

When the AC isolator is switched off, the Enecsys website shows power is being produced as normal. I switched off for 10 minutes to allow for any data processing delay.This shows that, even if the monitoring is done on the AC side, it does not measure the power delivered.
The Elster red light went off, indicating no reverse power flow.
Since the red light can be on under dull conditions, it tells me that the microinverters are actually taking some small power even at the same time that Enecsys says that the panels are producing power.
Also the Enecsys system can sometimes "see" all the panels without recording any production.

No reply yet from Elster but they are probably closed for the Xmas break.

What is now needed is some bright conditions to compare the power readings.

Wickhams' data indicates a loss of 5.9kWh in December compared to my loss of around 14kWh midNov-end Dec.
But I never get direct sun at this time of year so production is always low whereas I recall Wickhams site is pretty good.
Personally I favour the explanation that there is a static power loss in the micro-inverters but the jury is still out without more detailed data or any feedback from Elster or Enecsys.

[Wickham]

December on its own was 57.1 compared with 63 for Enecsys, a 9.36% reduction for a very dull month.

That would be consistent with the Enecsys monitor reading the DC input and the micro-inverter having an efficiency of ca 92% - reasonable.  Interesting thing is, jez54, if it is indeed so that Enecsys report the DC side (only), then it is awfully hard to identify a faulty inverter.  You might have to run this for 16 days, each day with a different inverter disconnected from its junction box (always comparing Enecsys monitor to generation meter).  PITA.

Surely that's easy as each micro-inverter is displayed separately on the website so you look for a different hourly or daily total for that unit, or none. That's how I found out that one of mine had failed (I got a free replacement after a delay as Enecsys had stopped all production from Poland at the time as the manufacturers had made unauthorised amendments).

[series530]

A bit more data from my system:

It's been consistently dull here today and my panels claim to be producing 14w each. That gives the 196W which is consistent with all of the Sanyo's being on and just producing power.

Enecsys says that the system production is 196W instantaneous and 196wH typically for most of the day.

The Envir claims to be seeing 67W through the system and about 0.63KwH on average today.

The AC100 is pulsing every 55 seconds giving 0.65KwH of production using the calculation of 100 pulses per KwH.

Thus, the Envir/Optismart seems to be consistent with the pulse rate from the AC100. Both, however, are not consistent with the Enecsys monitor.

HPSauce has, I think, given the vital clue : his system doesn't get any direct sun at this time of year. I know that mine does and I suspect that most other systems do. This is what I think is happening and may explain some of the inconsistencies seen between the systems:

I know that the Enecsys inverters get their power from the panels and not from the grid. My installer told me this on the first day when I rang him up in a blind panic to tell him that they had mysteriously all suddenly switched off. The question is if they monitor the power on the panel source side or the grid side or the inverter. Depending upon how you measure that data, this may or may not be significant. We'll come back to that a little later.

Lets consider the AC100 for a moment: as I stated in a previous post, the AC100 seems to have a starting current of about 100ma. So, any current flow of less than 100ma will not trigger the unit. How significant is 100ma? Well, on a 4Kw system which is outputting, say, 3.5Kw at 240v (and ignoring power factor), the current produced is about 14.6A. 100ma in that is less than 0.7%. Let's say that the system is producing a more modest 1.5Kw: that's 6.25a and the 100ma is 1.6%. What about 200W? The current now is 0.83A and the switch on current is 12%. I suspect that the current required to switch on the meter is simply ignored by the meter and any readings that we get are offset (against our favour) because of it. At 240V, we loose something like 24W just because of this. It's also possible that the meter gets its power from the PV line. We could loose a few watts more simply to run the internal measuring instrumentation, the protection circuits and the micro controller.

Let's go back to the Enecsys monitors again now. We may have lost 30W or so in the meter (due to the ib offset and the power to run it) but we are still down by about 100W according to my system. 100W, shared across 14 harvesters is 7W per inverter. The zigbee transmitters are known to be low power so I doubt that these are major contributors. Of the rest, maintaining an MPPT, constantly tracking for optimal power, sensing the grid supply.... could this consume 7W? quite possibly. With a 250W panel, a 7W sacrifice is about 3% of performance. This is a little better than the stated Euro efficiency but I doubt that the Euro efficiency is based upon a theoretical maximum capability (?). If we ran at 50% of stated maximum output, a drain of 7W on 150 would about 5% and more in line with published data.

So, why is HPSauce suffering more than the rest of us? If think the issue is the direct sun. The losses in pure power terms because of the inverters and the AC100 are relatively fixed numbers. For most of, even in the gloomy December that we have had, there have been some bursts of sun. Those bursts of sun have spiked our harvest for a few minutes or even a few hours. One hour at 1Kw would be the same as 10 hours at 100w. If HP was only ever producing 100W, he would be loosing most of that because of the base current of the Ac100. He would also be loosing a large portion of that because of the quiescent power drain from the inverters. When we have no direct sun, our efficiency is as bad as his. When the sun shines for us, we spike our harvest, the standing losses become minimal compared to the quiescent dull day harvest and this improves the overall system efficiency and, as long as we get a reasonable amount of direct sun, our efficiency will always seem to be much better than his.

... well, that's my theory (for today), anyway