Dynamic ESS Groene modus: PV teruglevering terwijl accu later vanuit net wordt geladen

Been noticing some odd behaviour with Dynamic ESS lately and wondering if anyone else has seen this.

Running a Victron setup at the cabin — Cerbo GX with Venus OS (kept fairly up to date), feeding into a MultiPlus-II. I've got it configured in what I'd call the "green priority" mode, where the system is supposed to favour solar self-consumption and avoid unnecessary grid interaction.

The issue I'm seeing: on decent solar days, the system is exporting PV to the grid whilst simultaneously scheduling a grid charge for the batteries later that evening. Feels completely backwards — why dump energy you've generated for pennies and then buy it back at a higher rate a few hours later?

My suspicion is the algorithm is doing some kind of price-based forecasting and deciding the sell-now / buy-later arbitrage makes sense economically, but it doesn't feel right when the whole point of the green mode is to minimise grid dependency, not play energy markets.

A few questions for the group:

• Has anyone else seen Dynamic ESS behave like this in green/self-consumption mode?
• Is there a known interaction between solar forecast data and the charge scheduling that causes this?
• Would tightening the minimum SOC settings actually prevent the grid top-up, or does Dynamic ESS override those?

Context: I'm also running EV charging off the same system, so battery management is pretty critical for me — I can't afford the batteries to be half-empty if the solar forecast turns out wrong.

@GemmaWood — seen exactly this with my island setup. Dynamic ESS in Green mode will happily export surplus PV whilst simultaneously scheduling a grid charge later in the same day if the algorithm decides battery SoC won't be sufficient for the evening peak window.

It's not a bug per se — it's the optimiser treating export revenue and overnight charging cost as separate decision variables. The issue is when the forecast is wrong (cloud rolls in, as it does here in Scotland), and you've already exported cheap afternoon energy only to buy expensive evening units back.

Worth checking your Min SoC floor setting — I bumped mine to 40% which forces the algorithm to prioritise battery retention before allowing export. Also verify your ENTSOE pricing data is actually pulling correctly through the Cerbo; I had a period where it was optimising against stale tariff data.

@GemmaWood this is basically Dynamic ESS doing what it's designed to do — it's optimising around the tariff schedule, not just raw self-consumption logic. In Green mode it'll export PV if the feed-in rate makes sense right now, then buy from the grid later when it predicts a cheap window. Whether that's actually beneficial depends entirely on your import/export rates.

Worth checking your Charge Schedule entries in the Cerbo — if you've got any manual overrides in there they can conflict with Dynamic ESS decisions and produce exactly this kind of head-scratching behaviour.

Also double-check your Venus OS version. There were some Dynamic ESS quirks that got patched in recent builds. What firmware are you actually running? The changelog on the Victron community site is surprisingly readable if you dig into it.

@GemmaWood I'd be a bit wary of assuming the optimiser is actually getting this right. In theory yes, it's arbitraging export vs cheap-rate charging, but the model relies on accurate price forecasts and decent PV predictions — both of which can be pretty ropey in practice.

What's your export rate versus your cheap-rate import? If the margin is thin, the algorithm might be "winning" by pennies whilst adding unnecessary charge cycles to your battery. Long-term that's not obviously a good trade on LiFePO4 cells.

I've not run Dynamic ESS myself (shepherd's hut, simpler setup), but I'd want to see actual data before concluding it's optimising rather than just... doing things. Pull the VRM logs and compare what it predicted against what actually happened. That'll tell you whether the behaviour is clever or just busy.

@GemmaWood this is something I ran into building out my tiny house system last year. The key thing worth checking is whether your Cerbo's time sync is solid — I had Dynamic ESS making genuinely baffling decisions that turned out to be because NTP was drifting and the optimiser was miscalculating where it sat in the tariff window.

Also worth pulling the VRM portal logs and overlaying your grid import/export against the tariff schedule manually. Once I did that, what looked like irrational behaviour was actually the algorithm correctly anticipating a cheap overnight window and deliberately running the battery lower beforehand to maximise that charging opportunity.

The optimiser's forecast horizon matters enormously here — if your PV forecast from the VRM cloud is off (overcast days are notorious), it'll make decisions that look wrong but were internally logical given the data it had.