You can get devices that determine the current irradiation, but they are not cheap.
So it has been discussed over years on how to determine the panel potential, to then switch off/on loads … and then a cloud moves over.
The challenge is not the good days, the challenge comes in on weather days.
Or get data from someone near one’s house: https://app.weathercloud.net/
And then interface that data with one’s system to switch loads on-off. But that gets really messy and not in one’s control.
Another way is to check what goes in/drawn out of the batteries … and use that as the point of reference. I use that to throttle the inverter on bad weather days, to recharge the batts, or at least, to not use them all the time.
Sometimes simple is beautiful. In my house there is one automation rule that says turn geyser off if overall consumption is over 5kW, turn it back on if under 1.8kW.
The difference between the points must be more than 3kW, otherwise it will oscillate.
It is far from perfect and it doesn’t even know if the thermostat is ON at any point, but it works. If the load is already very high, either turn off that geyser (or prevent it from switching on). If the load is low, allow geyser to be on.
(This is the second geyser, a 150 liter unit used for the hair salon next door. It can be switched off the whole day without danger of running out of hot water, the capacity is sufficient).
It doesn’t have the smarts to check if there is enough PV to switch on with zero grid use… but I don’t care. I still get around 45% of my energy from the grid anyway and a crude form of “load shedding” is simple and beautiful.
I have found that a small (couple of watts max.) solar panel with a resistor wired across it can give you a very nice indication of available solar power. You need to choose a resistor that would draw more power than the panel could produce in full sun (You should basically treat the solar panel as a current source, where the current is dependent on the irradiance). Then the voltage measured across the resistor will give you a pretty accurate idea of available solar power. You calibrate it by comparing the power that your PV system can produce to the voltage across the resistor to get a scaling factor. Just multiply the measured voltage across the resistor by your scaling factor to get available PV.
I use a weather node in Node Red to access an UVI sensor on a local WUnderground weather station. This gives the UV value which I can then use to start and stop various bits.
For the geyser (2kw element) its more simple (Sonoff R2) being used to switch geyser (it has a timeswitch if needed as well).
SOC greater than 80% = Geyser is ON
SOC less than 70% = Geyser is OFF
What (still) puzzles me is why the PV power isn’t available as a field in the inverter (or controller)
The grid tie inverter has to back off excess PV power to prevent powering back into the grid. (Assume the system doesn’t have any storage)
Does it simply detect the power being fed back and the controller then backs off the MPPT until this is close to zero??
Far as I know, yes, hence on Victron you have to use AC_out1 and AC_out2 or use a Carlo instead of AC_out2, either method giving the software the reference point of what is being pushed back, or drawn.
Grid-tied inverters also use those current clamps to see what is going on on the AC side between the inverter and street.
The inverter is controlled so that it feeds as much power into the grid as is required to get the reference point (either the active AC input, or the grid meter) to zero (or whatever you chose as your setpoint, default is 50W). It will draw whatever power is necessary to do this from the DC side.
Let’s assume the batteries are full at this point for the rest of the illustration. That means your MPPTs are in voltage/current limit mode (not in MPP mode).
If it draws less power from the DC side than the PV can provide, the battery remains at the configured charge voltage, and the MPPT simply ramps up to provide the required power. It is voltage controlled. As power is taken from the DC bus, the MPPT ramps up to keep the voltage up.
If you use enough power from the DC bus, the MPPT may decide to switch back to MPP mode. It will do three quick scans, and then pick the maximum power point.
The MPPT doesn’t know where the maximum power point is while it is in voltage/current limit mode. It is running on the backside of the P/V curve, and it may have been there for hours, so even if it had previously known where the MPP is… it might have moved.
Hence, nothing in the system knows where the MPP is going to be until it has to go out looking for it. And as such, it cannot predict available PV while it is not in MPP mode.
SMA advertised a feature some time ago, where the MPPT would remember a previous max power point and revert to it as a starting point should it need to find the maximum quickly. I don’t think they use it in a predictive manner, and again, that is useless if your system has been in voltage/current limit mode for the last hour. Solar conditions have changed. You don’t know where the MPP is.
Edit: Just to make the initial part clearer. There is never a command that goes to an MPPT that says “please send me 1000W”. There is a command that goes to the MPPT saying “please make your charge voltage 56.8V”. And then the MPPT will do that… it will do whatever it needs to do to keep the voltage at that level. The AC/DC conversion side then does its own thing… and if you command enough power, this automatically causes the voltage to drop, the MPPTs to kick in, etc etc. Voltage control. Not current
Aah So!
So it’s only on the AC side that current (power) is measured…
And in regular grid tie mode there is no limiting of the PV power and this will be drawn by your load and the excess fed back to the grid (heaven forbid!)
Way back I engaged Youda (remember him?) on this subject but somehow he didn’t manage to get this concept to penetrate my (thick) skull at the time but I’m sure it helped…
Ok: So now to do some more thinking…
Yes, but I think the point here is that once you are limiting feedback on a grid-tied inverter or in battery voltage control mode on a DC MPPT, it is impossible to know how much PV is theoretically available that you are not using. Unless you measure solar irradiance.
This is a very good point and has caught me out a few times. When battery is full and loads just ticking along and you switch on geyser remotely thinking there is plenty sun only to see that there is nothing extra from the solar to run it and it chows your battery.
Indeed! That’s why your idea of a separate PV panel (with load) will provide an indication of sun power.
Measuring power on a MPPT with it’s varying impedance has got to be tricky!
Of course, if you don’t mind a little bit of grid import, you can always just switch the load and see what happens. Unless you are already grid neutral or import very little from the grid, there is no reason why this cannot work. Who cares if you import a little just to do the check… you already import energy anyway.
That is my case. I import about 50% of my Energy from the grid. It was intended to be this way from the beginning. If there is a grid outage, I want most of the usual stuff to continue running, but not everything. If I ever end up in a situation where I’m importing less than 5kWh a day and want to lower it even further, THEN I will look at fancy ways of measuring the available PV
My plan (I have been planning to do this for a year and have yet to get to it) was to use the battery to measure the available power by starting the geyser when the battery has charged up to about 90% SOC. This would be before the PV get throttled and I can just read the PV power as what is available. It will only work for starting the geyser and not if the PV changes after it starts, but then by geyser(heat pump) only need 45min @900W to heat the water.
(Edit: the small PV panel is a much better option)
This isn’t quite what happens. I asked Kwikot this question and this is their response:
The water will behave in a thermal cylce when the element is on – as soon as it switches off the water stratifies into layers with the hottest on top. The thermostat will then switch on again until this process has evenly distributed the stratification.
If you feel the inlet pipe of a geyser that has been on temperature for some time you see that it is hot.
At least 80% of the water will be at the thermostat setpoint after say 6 hours.
That is correct. But for the situation where you heat your geyser until the first time you measure the setpoint, and then switching it off, you have only heated half the water. Then it will start mixing, but that process is surprisingly slow - if you have a Class B geyser you should be able to test it by heating the geyser and plotting the temperature at the flange probe point. You should see an exponential decay with one slope, and then the slope will decrease dramatically once the temperature is at a homogeneous temperature.
