This started happening after upgrading from v.3.80~44 to v3.80.~45.
So the issue is that when sell price is high, DESS in trade mode decides to feed to grid rather than charging the batteries. When price drops, it switches to battery charging, and this is where the problem occurs. All of the PV power (e.g. 3kW) go to batteries, but at the same time self-consumption (e.g. 200W) is taken from the grid although there is plenty of spare energy. It appears that the inverter is deactivated at the switchover point. Workaround is to briefly switch to “Keep batteries charged” mode and then back to “Optimized with/without BatteryLife”. Regardless if it is a new problem related to version ~45 or it existed before, how to diagnose it? Is it possible to see instructions GX device gives to inverters (Multi RS Solar in 3 phase configuration)?
There is seldom anything in the base venus OS or beta changes that affect DESS, with it being driven from VRM, not the GX.
To be sure, you need to rollback to a production release (3.79) and see if it persists.
It is typically a coincidence and unrelated.
Also worth considering is if nodered and virtual devices are in use and if anything could affect those flows.
Because it takes a long time to reproduce, I did not switch between versions, but because DESS is controlled by VRM, let’s assume the problem was not introduced in 3.80. I just did not notice it before.
I took some time to familiarize myself with how DESS works and I now understand that there are cases where using power from the grid while there is solar production is sometimes the better approach than using solar for self-consumption. This is when the buy price is lower than the sell price in the evening, so selling solar later is more beneficial than using it to compensate for self-consumption. But I am now also convinced that there is a bug in DESS or Venus OS after all, where the inverter is not put into the correct state. My guess is that if during the night the grid is used for consumption, inverters do not get ordered to start compensating for self-consumption and stay deactivated when solar production starts. DESS schedule showed that solar will be used for consumption but it did not happen.
As you can see in this picture, the sell price in the evening is lower than the buy price during which DESS was not using solar for self-consumption. So I did my workaround of switching the mode to “Keep batteries charged” for a second and then back to “Optimized without BatteryLife”, and then the inverters kick in. You can clearly see where Grid to Consumption stopped and Solar to Consumption started. That is the point where I did the trick with switching ESS modes.
I was watching DESS parameters before and after my trick but nothing changed. Used strategy was Self-consumption, Feed-in enabled and active restriction was - Grid to battery energy flow is restricted. What else can I watch when this happens the next time?
Same thing this morning. All night was a “grid to consumption” because price was high the previous evening and all of the battery was discharged. DESS had scheduled to charge the batteries as soon as the sun comes up, but it forgets to turn on the inverters, so I continue to receive power from the grid. This time I installed VenusOS 3.81 at the point marked with arrow after which all goes to normal, so I must either briefly switch to “Keep batteries charged” or reboot the GX device to solve the issue.
I see the same on 26 Sept and this morning: after a night of idling, the inverters stay unable to discharge until you toggle the mode or reboot.
Next time it happens, before toggling, could you check in Remote Console (or dbus-spy) on com.victronenergy.acsystem the values of /Ess/DisableDischarge and /Ess/DisableCharge? Also, does any Node-RED flow write to ESS settings?
This morning was not supposed to be the time the problem occurred, as all night was supposed to be set to “Battery to consumption”. However, due to the problem I reported here, DESS always leaves too little charge in the batteries, causing them to run out in the middle of the night. DESS is probably not accounting for inverter consumption when they are left active. You can see that the batteries are even being charged slightly to prevent them from dropping below the minimum SoC when the inverters switched to consuming from the grid.
So I checked the values you asked for: both DisableDischarge and DisableCharge were 0, however DisableFeedIn was 1. When it is reset to 0, everything returns to normal. The batteries are then charged, and grid power is kept at 0W.
And no, I am not writing to ESS using Node-RED. Currently, I am using it for virtual sensors and outputting some values to Dashboard 2.0.
Thanks, that helps. DisableFeedIn = 1 is set by DESS itself, because last night’s schedule didn’t allow feed-in until 08:30. But it’s probably not the cause on its own: between 09:00 and 09:15 it was 1 again and your loads stayed on solar, and on 9 October it was already 0 from 08:15 and only the reboot helped.
While it’s stuck, the RS passes grid power through to the loads and all PV goes to the battery. Your grid power dropped to 0 at 08:33.
At what time did you set DisableFeedIn back to 0?
Next time, before changing anything, could you check in VictronConnect whether the RS shows inverting or passthrough, and any warnings? And do you have a separate grid meter?
Yesterday it started working by itself without me setting DisableFeedIn to 0. While I was making screenshots and posting here, I noticed that DisableFeedIn changed to 0 and everything started working fine. I think this happened because the batteries charged faster than predicted, so DESS decided to discharge the batteries a bit, which brought the system back to normal.
This, by the way, is yet another problem with DESS which I have noticed before. If solar generation exceeds the forecast by a large margin, DESS may end up feeding back to the grid the energy it has just stored in the batteries, all within the same 15-minute slot. It probably tries to get rid of excess SoC by discharging the batteries to the grid, which is not the right decision in most cases.
No, the internal meter is used as the whole house is connected to AC Out. I will check what RS is showing when this happens next time. I am not sure when this will be, as it requires using the grid instead of the batteries at night and then charging from the sun in the morning (expensive evening price and then cheap night and morning price).
