OpenEnergyMonitor Community

Solar PV Divert

I have begun using solar PV Divert. I’ve been tinkering with the smoothing and decay adjustments since I noticed that on days like today where it’s cloudy, foggy, misty, etc and output is a bit all over that the divert feature will spike.

On the right I have managed to tap it down a bit.

I was wondering if something that could be added might be a maximum current increase/decrease. We still have the smoothing and decay adjustment, but also have a hard stop where we could say increase of 3A between EV adjustments.

One of the things I’ve noticed is the EVSE will adjust the pulse, but my EV is slow on adjusting to the new pulse. This causes the EVSE to adjust a number of times overshooting the excess PV output and then to correct that slamming itself down to 6A. It yo-yos a bit before settling. I think hard stops in between the EV change might help prevent that. Make it more step like especially on days like today here.

The smoothed values actually only dictate when the charging is starts/stops, the actual rate is directly attached to the available solar, the theory is that changing the charge rate is a lot less stress on the EVSE/EV whereas starting/stopping the charge can cause additional mechanical ware on the relay.

If that assertion is not true then maybe what would be best is to add additional attack/decay values for the charge rate so that can react faster but still smoothed.

I was not so much referring to stopping and started the charge which would cycle the contactors. I was referring more to the charger adjusting the pulse sent setting the charge rate. Following it I noticed that it would adjust the rate steps a lot though the EV had not yet adjusted to the new rate. So then when the sample was taken again the charger would see more available power and adjust again. By the time the EV adjusted it’s charge rate to match the new rate set by the charger it would overshoot the excess available and then the charger would plummet the rate down. You can see that in the attached screenshot.

My thinking was an added option to set the up and down current limit between EV adjustments. Say allow a user to set a hard limit of 5A between EV adjustments. So, we have excess of 2000W and currently the EV is at 6A. The EVSE can adjust to take up that excess, but only to 11A. Then the EVSE must wait for the EV pull to be 11A (or realistically 10.58A or whatever it’s pull is). Then let’s just say there is still 1500W available. The EVSE can then set 16A. This will repeat over until the set excess is taken up, but it will happen slower to help prevent the massive swings I was seeing in that screen shot. Trying to dampen the yo-yo affect.

Sure, that is effectively what you would get with additional smoothing settings too. The problem with fixed values is that they are very case specific, so while they help in one case they would not be as efficient in other cases. This is something we learned very early on when looking at the divert code improvements.By smoothing out the values you handle a lot more cases (but by no means all).

@glyn.hudson Maybe we should try to get the simulator integrated as an app in EmonCMS so folks have a tool to experiment with different values.

… although I almost certainly have broken the simulator in v4 :frowning:

That was why I figured those fixed values would be user selective. I choose 5A. Someone else chooses 1A, etc etc.

Unless that is also to rigid and narrow. Basically, like that it can do this, just want to smooth it out and stop the large swings. So, figure we can have a discussion.

I might also be where instead, just have the EVSE not make repeated changes while it waits for the EV to adjust. That might help, I saw it up the current limit repeatedly when more PV excess was available and the EV had not even changed to the first change.

The fixed values will be a problem for the same system on different days, for example say you set it to only change by 1A to handle days that are very spiky, on days there you have no clouds or infrequent clouds this is not going to be very efficient and you would probably find yourself tweaking the setup multiple times a day to get the best out of it. By smoothing the solar it handles the changing conditions better.

Actually you do bring up a good point that we do know what the EV is consuming so we do have a feedback loop, so maybe a PID controller will help. Something along the lines of Using the PID to modify the smoothing values based on how quickly the EV is responding and the rate at which the solar is changing…

I was wondering if you did take in to account what the EV was pulling at the time of calculation. Doing so would certainly help the controller smooth it out. At least then the EVSE might not, for example, change the charge rate pulse repeatedly before the EVSE has had a chance to change. That might at least slow it down and get better data for the next calculation. Plus, it also would allow it to calculate what the EVSE is pulling in power, match it to the available excess, calculate the power available for the next step, calc the amps change and then add it to the step.