I’m currently using Dynamic ESS (DESS) and noticing a gap in logic between Green Mode and Trade Mode regarding self-consumption and price optimization.
Current Issues:
1. Trade Mode vs. Self-Consumption Preference
I would like an option in Trade Mode where self-consumption takes priority over active trading, but price optimization still applies.
Green Mode aims for self-consumption, but it starts charging the battery immediately in the morning using direct PV power.
In dynamic pricing markets, morning and evening prices are often peak/high, while midday prices drop (or go negative).
Charging early prevents selling high-value PV power in the morning when it would be far more profitable to sell PV early and charge the battery during cheaper midday hours.
2. 48h Horizon Planning Before 14:00 Day-Ahead Price Publication
Currently, before new day-ahead prices arrive at 14:00, DESS seens to lack sufficient forward planning for the next day’s domestic load and PV generation.
To guarantee maximum self-consumption and optimal battery readiness for the following morning, DESS should estimate/project the next day’s solar yield and consumption profile even before the official 14:00 dynamic tariff release.
Proposed Improvement:
1. A configuration toggle or target in Trade Mode: “Prioritize Self-Consumption” or just a new called “Hybrid Mode”.
2. A rolling 48 hour yield/consumption forecast buffer that operates independently of the exact 14:00 tariff update window when self consuming has the highest priority.
At first it seems counter intuitive, but if you can sell electricity to grid, self-consumption prioritization will decrease your earnings. If electricity is cheap currently, but will get more expensive later, batteries will end up with less charge when expensive hours come if you self-consume during cheap hours. For example you buy several kWh from the grid for a few cents even though there is solar production. Batteries will have those several kWh more to sell at a much higher price if you did not prioritize self-consumption. You having to buy electricity from the grid is compensated by larger earnings later.
That is exactly what DESS does in trade mode. It might even not charge at all if price does not get more expensive in the evening or during the night.
I concur with @Slowmo that more self consumption will probably lower profit.
And I don’t think Victron will revisit the long-standing issue of DESS Trade bring sell biased towards minimum SoC at the end of the known pricing window, a search on my handle will show I’ve been persistent in trying to a point I decided to stop just log out and stop participating on this corporate forum.
That said, this being my so called pet peeve issue, I can’t resist making an exception to share how I solved it, almost doubling my trading results. It’s easier just to tell you how to do it, it will be self explanatory once you see it in action:
Check your DESS grid and battery power limit settings and make sure the battery discharge limit is (much) higher than the grid feed-in limit
For a limited time until you know what works best for you:
Everyday around 14:00, after new prices arrived, manually edit your forecast in VRM:
Set at the end of day tomorrow, an EV forecast of half your battery capacity
Reset/remove the EV forecast that was set yesterday, at the of (now) today
Wait 15 minutes and check your SoC forecast, adjust the EV forecast to your liking
After some days, when you are satisfied with a good EV forecast value, automate the dayly roll-over in Node-RED
The above will already provide most of the achievable improvement, once automated it is up to you whether you’d like to squeeze the last bit of extra profits as well by implementing a smart(er) EV forecast value. PS, you can also use the other forecast fields but I am accustomed using the EV field.
I made something in nodered already for switching between green and trade mode depending on the consumption and solar forecast, it is related to the fact trade mode doesn’t plan for 48 hours which causes grid buy the next day when the battery starts charging to late.
Self consumption will indeed generate lower profit, but that’s the same for green mode. But like I said, green mode starts charging in the morning while it could wait for noon when the sell price is low.
What I can do is use trade mode but add 10 cents to the sell price formula, this will prioritize self consumption from battery but it has 2 trade-offs. 1. The calculated costs and profit will not match reality and 2. the battery won’t charge from grid when the real price is negative.
I don’t want to manipulate dess to much with nodered, because I also want to give feedback to Victron so they can improve Dess which hopefully they will.
You don’t have to believe me that my method/workaround solved the actual root cause of the issue, you can test it and see for yourself. All I can say is that there are secondary effects that are apparently hard to convince people of verbally, but will become trivially obvious once you see it in action. Last half year on average €150 profit/month with MP-II 5000 on 40A mains (but with boost charger added), no solar, low household use
Preamble: by now tomorrow’s pricing is known, therefore you should set your visible time period to: ‘Next2 days’ and roll-over that EV forecast to end of day tomorrow.
That said: yes and because the EV forecast gets rolled over to the next day before we arrive at the forecast time, DESS actually never arrives at the forecast. This way the forecast becomes a means to an end: to inform the DESS scheduler to reserve a certain amount of energy at the end of the known pricing window. And has in practice the same effect as having a longer estimated forecast window as you were thinking about. Basically you simulate the expected next morning sell opportunity, and suddenly the DESS Trade algorithm starts producing a pretty well optimized SoC plan. Key point herein is never to reach the forecast slot at end of tomorrow, nor today.
My beef with the devs in this matter, concerned the difficulties we had conceptualizing why this was needed in the first place. My stance was and is, that it an architecture issue, that the minimumSoC value is used for two distinct different functions: the ‘actual’ battery minimumSoC, and the ‘target’ planningSoC value you wish to plan for at the end of the known pricing window. The real solution is to separate the two.
So for what I understand is that by planning EV charge (or any other consumption) at the end of the day ,which is removed after the planning for the next day is ready which raises the soc at the end of the day, the battery is charged a bit earlier in the morning so it has more energy to sell during the evening and the morning next day and that creates extra profit?
Just as a side note, I’ve been writing about this for quite some time. Extending the forecast period, whatever weaknesses it might have, would account for periods of bad weather and, in my view, could lead to much better coverage of self-consumption.
