Guten Abend, ich habe seit ein par Wochen das Problem, dass DESS keine Grundlast mehr einplant.
Das bedeutet, wenn Abends die Batterie wegen hohen Preisen entladen wird, die ganze Nacht Netzbezug ensteht weil der mindest Soc zu früh erreicht wird.
Hat jemand die selbe Beobachtung gemacht?
Hi @LukasK1 ,
That is unfortunate! Did you try out the new ‘adjust consumption forecast’ feature?
I checked out your site and I see that DESS does show the consumption forecast, and following a similar consumption pattern as your last couple of days.
Hope adjusting the consumption forecast for yourself helps, and eventually the model will learn from that too.
Hello Barbara, yes i did that for one day and this works as expected. it’s just not a solution for my problem. i can’t put in so much values every day again. should i try to delete the learned values of dynamic ess to fix the problem?
Hi @LukasK1 I don’t think you need to do anything right now, you are correct: I was looking at the total consumption vs forecast, and that seems correct, but DESS assumes you do not consume once the sun is down.
I’ll pick it up with the team and come back to you!
thank you very much
Hi @Barbara, did you already fix anything? it looks like my forecast after sun is down works again.
Same (or similar) problem here. There are days where DESS calculates nighttime consumption as zero (as example 0.0kWh from midnight till 5am instead of ~0.1-0.2kWh per hour) which causes Trade Mode to discharge the battery too early. It might be related to high loads during the day, I mostly noticed it on weekends where I charged my car before. Quick fix was adjusting forecast but it would be great to have a separate Trade SOC setting so that there is some reserve for situations like this.
Hello @Barbara,
I can also confirm this phenomena. I have troubles already several weeks with very low quality of the forecast (factor ~3-4 at some periods of day and I have also troubles with the zero prediction of base load during night.
Forecast :
Reality - my night base load is stable over month-years, ~0,5kW, never 0, very stable during all year. Time to time there can be car charging, but charger is Victron and visible as EV charging in VRM not within base load:
Typical day:
I have to daily modify my forecast now hour by hour. I assume that this problem started with the beginning of high PV production at spring + with the negative grid costs.
That time the loads (as I use huge water boiler + EV charging) were changing between 20-150kWh daily. Model changed forecast quite fast and start to predict immediately very high loads. Now when the consumption reduced it seems that some reduction factor is used over complete day inc. night…?
Influence of one-two special outliner days seems to be too hight on the forecast and when returned to standard → it takes weeks to return prediction to normal.
Again → outliners has to be filtered from the model and the same for EV and heat pump. Only base load can be predicted well. EV charging and heat pump cannot to predicted mainly based on history but more by weather forecast, PV production or better controlled by DESS directly.
@Barbara Hello Barbara, could you figure out any problems until now? i have daily grid costs because of this problem. i cant even set the consumption by hand to solve this problem
