question

As a follow-up to my previous post, here's the forecast for today. No overnight charging at low rate, then peak rate electricity usage in the evening. And even a short period of charging the battery during peak cost. I am trying very hard to see the logic - financial or otherwise - in this algorithm. Far from saving money, DESS seems to be doing the opposite. Target SOC seems to be way too low, and there seems to be no provision for uncertainty or probability of the forecasts.

Can anyone from Victron perhaps comment on why the forecasting is so far out from what would seem to be sensible?

The algorithm was expecting the solar energy to be enough for the night, but it turned out it wasn't. Since Dynamic ESS is need-based it simply leads to not charging ‘for no reason’ during the low price energy hours.

If you compare your first image to the actual one, you see that the solar yield was a lot less than anticipated.

In hindsight the system should have charged during the cheap hours. But hindsight logic is always easy and you can't change te past.

hi @Dirk-Jan Faber (Victron Energy) , thanks for the reply. I understand your point.

I'm in the UK and we are blessed with unreliable solar and very changeable weather as you know. Forecasts are good and bad - or rather high certainty or less ...

My point was regarding the uncertainty of the forecasts being used. There are two uncertainties that are independent of each other - the solar forecast and my consumption forecast. A meteorological forecast usual has a probability on accuracy. When it comes to my consumption, you have no chance ... history is ok, but if I decide on having a party tomorrow, the forecast will be out. So, how should those uncertainties be built into the calculation of target SOC? I am assuming of course that they are not!

Wouldn't it be possible to include some degree of flex on the target SOC based on the solar forecast probability? The more uncertain the forecast the higher the margin added to target SOC.

Own consumption is more of a problem, as only I know what I'm doing tomorrow. But maybe a simple calendar recording planned usage criteria. The user could specify types of planned usage, e.g. Vacation (low use), Normal (normal use), High (like Christmas Day) for specific days. This could be used in a similar principle to adjust the consumption forecast and the target SOC.

On a related point, if you turn ESS off temporarily does it (i) remember the history, or (ii) start accumulating history from scratch when you turn it back on?

IMO (i) would be better, as you could then turn off DESS when you go away as there less need for it, and turn it back on when you get back home.

What does it actually do in practice?