Three phase Consumption and Three phase Generation


I've got a 3 phase building that we've just had 3 phase solar installed onto. The inverter doesn't support PVoutput (only via CSV) and I would love to be a part of that so I'm looking at Flukso coming to the rescue.

I would like to measure both our buildings consumption as well as the solars generation.

Would I be best off using TWO fluksometers (6 clamps in total)? If so how would that work as an input to PVoutput? Would I end up with 6 graph lines?

Thank you

michi's picture

CSV export isn't such a bad idea. If you can a Raspberry or some such, you could set that up to upload to PVOutput. I'm still doing just that for my SMA Sunny Boy 5000 TL-20. The Raspberry reboots itself at 3:00 am every day (just in case), and it's been rock-solid for something like 18 months. I'm using sbfspot, which has been working flawlessly. There are a number of other monitoring projects you could use.

With this, all you need is a Raspberry Pi with a bluetooth and wifi dongle. The Raspberry Pi can upload directly to PVOutput via sbfspot, so you'll need a Flukso only for monitoring consumption. (That's what I'm doing.) Because the data comes straight from the inverter (not measured by current clamps), it is dead-accurate.

I strongly recommend to invest in a 3-phase meter with pulse outputs. You will get *way* more accurate consumption data with a proper meter than with current clamps.



gebhardm's picture

To second Michi: I assume the "three phase inverter output" is a balanced one, that is the phase supply is (nearly) equal on all three phases. This would imply that you measure PV output with "just" one three-phase pulse meter which interfaces to a single pulse port of one Fluksometer. So just one FLMv2 would be needed.
The FLMv3 will provide three clamp ports and three pulse ports directly. This would allow you to measure PV output also with three single-phase pulse meters - so still just one FLM needed.
In contrast to Michi's SB5000TL-20 my SB3300TLHC is not very accurate, thus I recommend the pulse meter variant (if you are not a tinkerer who interfaces by bluetooth and makes everything in software anyway).

michi's picture

Just to be clear: I'm using a pulse meter to measure consumption (extremely accurate), and I'm using my Sunny Boy bluetooth link to measure production (also extremely accurate). No current clamps involved in my setup at all. My pulse meter is three-phase, but has a single pulse output, just as Gebhard describes. The meter provides aggregate consumption across all three phases. So, yes, a single FLMv2 is enough. (That's what I'm using.)

If you do get a pulse meter installed (highly recommended), I suggest getting one with 1000 pulses/kWh or more. You'll get better real-time readings that way, especially at times when consumption is low.

For consumption, a Raspberry Pi plus sbfspot (or similar software) is the way to go, in my opinion. Your production figures will be extremely accurate if you can read them directly out of your inverter. No headaches with phantom loads getting picked up by current clamps. In my case, the inverter feeds into a single phase of my three-phase supply, not into all three phases. But, regardless of what your inverter does, if it tells you that it has produced 1.00 kWh (whether it's been fed into your house or into the grid), it's likely that this figure is dead-accurate, irrespective of the number of phases.

Just make sure that you get your consumption meter wired in such that it measures total consumption (disregarding any offset from your solar system). That way, you get two figures, one for total consumption and one for total production, and the difference tells you what you have been spending/saving.



Enwise's picture

Thanks for the comprehensive replies guys.

A bit more info on my setup, it's zero export. I have a energy retailer compliant box that measures my consumption and ramps the inverter up or down to cancel out the load so that it equals close to 0 on each phase. So my generation graph would track with my consumption similar to the below.

Incidentally, that example above is exactly what I want to replicate with my system.

This box has 6 inputs (3 for each direction) and does it's own trickery inside a sealed unit (sealed by energy retailer). That's how I got to the idea of using 6 clamps in total over two units at this location. My consumption isn't balanced however I've not checked if the output while ramping is still balanced from the inverter.

With regards to a pulse meter, are you referring to something like the below? If so, I don't see a massive difference between this and doing it via split clamps? As long as the installation of the clamps is good they should be fairly accurate (I'm not trying to get billing quality data, more historical/trend orientated), after all the below is just the same thing but in a closed unit. I have a Fluke polyphase data logger ($3500 worth) that is essentially the same as the Fluksometer.

The Raspberry Pi angle seems interesting, but I feel it may be beyond my abilities after looking into it. I'm an electrician by trade, but computer building is not a strong field for me :D

Also my inverter is a ABB PVI 20.0 OUTD.

michi's picture

Interesting graph. I don't know why someone would want to prevent export of excess capacity and make the inverter ramp down to track consumption. What's the motivation for this? If there is excess production, you might as well export it to the grid?

Yes, I was referring to a three-phase pulse meter that sits in your meter box on a DIN rail. (The link you pasted doesn't work, but Schnap sell these meters; I bought mine from them too.)

Current clamps will give you a rough idea of current flow, but are not all that accurate, especially if you have loads with a power factor. Motors, fluorescent lights, and switching power supplies all generate phantom current that will be incorrectly measured as real consumption or export by clamps, whereas a proper meter will ignore them. Current clamps can read way high (as in 10% or more) depending on what loads you have on your circuits.



Enwise's picture

Crappy part of the approval process for systems over 5kW in our area (and in some cases under), we can't export any excess power which is unfortunate since at the moment we're using half our capacity (winter, no A/C on yet).

Right so I could still do the flukso solution using one meter with three clamps on the inverter outputs and one pulse sensor on a din mounted 3 phase meter. The Schnap meter I tried to link above is a 400 flashes per kWh unit.


michi's picture

Man, that's ridiculous. So you have excess capacity and are forced to throw it way? That's criminal :(

I'd recommend not to buy that particular Schnap meter. 400 pulses per kWh is rather coarse, and you'll be better off with a meter that can do 1000 or 2000 pulses per kWh. See this thread for some alternatives:

If you are not interested in real-time readings, then the 400 pulse meter is fine. It's just as accurate, just doesn't have higher resolution that you need to measure low consumption in real time.

Three current clamps on the inverter plus a pulse meter will work with an FLMv2.



Enwise's picture

Yeah, the party line is they're only allowing a certain % per transformer to be fed back in due to infrastructure limitations. The majority (but not all) of residential up to 5kW export up to that as a max, but larger systems are generally all zero export.

Those ABB meters look great, just no 100 amp version which is a pain. Our energy retailer meter does have an LED output, I've just been trying to find out details on it. It's a EDMI MK10A, but the retailer can set the pulse to what ever they want (usually 100 imp/kWh apparently for our region).

I'll keep searching for a 100a, 3 phase meter with a good rate. If not, that Schnap is rated to 100a so that'll do, especially for the price.

Thanks for the help.