Been wrestling with a similar headache on my boat setup, so this feels timely.
Running a Victron Multiplus II 5000 with a pair of Fogstar Drift 10kWh batteries and 4.8kWc of panels across two SmartSolar MPPT 250/70s. Mostly off-grid but I do have a shoreside hookup for when I'm moored in a marina.
The DESS (Dynamic ESS) scheduling has been brilliant overall, but I've noticed it occasionally pulls from the grid during what I'd consider peak-rate windows — even when the batteries are sitting at 70-80% and solar is actively generating. Genuinely puzzling behaviour.
My theory is that DESS is predicting a high-draw period later in the day and trying to top up "just in case," but the forecasting logic doesn't seem to account for real-time generation properly. The VRM portal shows the decision-making history, but decoding why it made a specific call at a specific moment is still a bit opaque.
A few things I've tried:
- Tightening the minimum SOC buffer settings
- Manually overriding the grid setpoint schedule in certain time windows
- Checking the BMS comms are actually feeding accurate data back to the Cerbo
That last one made a noticeable difference actually — turns out my Cerbo was occasionally losing the BMS handshake which may have caused the Multiplus to assume worse battery state than reality.
Has anyone else dug deep into DESS behaviour with larger battery banks? Curious whether the 16kWh NKON setup mentioned elsewhere in this kind of scenario behaves differently — bigger capacity should give the algorithm more confidence, but does it?