I have a programmable clock that starts my tapwater heating every day at 11.00 hour. The heating takes about 45 minutes and is about 7 kWh of energy. If I look to the consumption forecast than I see daily a forecasted consumption of about 7 kWh that is between 10.00 hour and 10.59 hour. That is one hour to early forecasted. See picture below. Is this a bug or do I have to change a setting?
1 Like
The next day I see the same. The consumption forecast is between 10.00 and 10.59 hour, but the real consumption is between 11.00 and 11.59 hour.
It is every day one hour difference between the forecast and the reality. See picture below. I think this is a bug in the forecast. Can someone react on it.
1 Like
Hello!
I have the same issue in my installation!
My water heater is scheduled to work from 5:00 to 7:45 and from 12:30 to 19:00.
In the 1st picture we can see consumption forecast that shifted 1 hour back compared to actual consumption:
- 4-5 instead of 5-6
- 11-12 instead of 12-13
- 17-18 instead of 18-19
And the typical actual consumption is shown in the 2nd picture.
Also I have tried to “Reset installation consumption forecasts” in VRM more then 3 times (last one was on 12 Sep).
This VRM feature does not work in this case! This bad forecast is disappeared for 3-4 days and after that is shown again.
So, it looks like issue with forecasting but not with this VRM feature.
As we approach deep autumn and winter, when the amount of sunlight decreases significantly, this will lead to incorrect DEES operation and as a result, overuse of the grid during hours when it is quite expensive!
Dear Victron support, please pay your attention on this issue, cause it is not unique and is reproducible on many installation, I am sure.
—————————————-
My installation:
3 x MultiPlus-II 3000 48 in 3 phase system
1 x MultiPlus-II 3000 48 via MK3/USB as genny charger with Smart Shunt on DC
Cerbo GX with Venus OS 3.65
Smart Solar 250/100 Tr VE.Can
Smart Solar 250/85 Tr VE.Can
Smart Solar 250/60 Tr
VM-3P75CT on the grid
Lynx Shunt on a battery
Seems I may have this issue as well.
My off peak period begins at 20:00. Last Friday we charged our EV starting 20:00.
Tomorrow (Friday) the projected consumption jumps one hour early (19:00).
@dognose or @Barbara could you possibly have the forecasting team look into this? Thanks!
1 Like
Hey! We are looking into this issue at the moment. I’ll let you know of our findings later this week.
2 Likes
Thank you! Not to go off topic, but it would be amazing if we could somehow exclude certain measured loads from the consumption forecast. In my example above, the EV charge totally (and correctly) throws off my consumption forecast.
It would be cool if we could add a separate VM3 meter to measure the EV charger, and somehow tell VRM to “exclude” this measured load from the consumption forecast (i.e., since we don’t charge consistently, so it’s almost detrimental to have DESS plan around it).
1 Like
Thanks, the issue is consistently present every forecast.
Hey!
If you use a Victron EV Charger or an energy meter with role EVCS, we already forecast consumption of the EVCS separately, following another model that is better at predicting EV Charger consumption.
2 Likes
Interesting! I’ll buy another VM3 meter and try it out.
1 Like
Hi @koningkronenberg ,
We have made an adjustment in the consumption forecast model, and only activated it for your installation for testing. Could you let us know if you see an improvement?
Then we’ll roll out the change to all installations.
I am also ready to test adjusted forecasting, so you may apply the changes to my installation too.
Ok thanks. I will test the adjustment and will update you with the test results.
Dear Barbara,
This morning about 8.00 o’clock I made a screenshot of the forecast. See picture below
The forecast is 6.1 kWh between 11.00-12.00 hour. The forecasted time slot is ok now.
The real consumption is 8.43 kWh between 11.00 and 12.00 hour. see picture below. So the timing of the forecast is ok now but the forecasted cosumption is a bit low. Normally I have a consumption of 8-8.5 kWh (but the case is about the timing of the forecasted consumption)
I will monitor coming days if I see some side effects of the solution.
Best Regards
Johan Koning.
1 Like
@koningkronenberg how’s the change been working for you?
1 Like
Hi A-P,
You can find my testresults in this thread just above. The timing of the forecast is working correct now. I see no negative side effects until now. I expect that Vicrtron will release this update in the next release, but that is not at me to decide.
Thanks! Glad to hear it’s been working as expected.
@Barbara thanks for resolving this. When we can we hope it is rolled out more widely?
I will check the coming days if it also works fine with quarter prices. That will start tomorrow.
We have rolled out the change to all installations!
3 Likes
Thanks @Barbara!
Now everything looks good in tomorrow’s forecast, will check how it works.
1 Like
Dear Barbara,
Thats great to hear, i was looking into categorising my load with meters. What kind of loads could lead to better forecasting? If i for example add a meter for my heatpump, will the forecast consider the weather for it?
What could be a great feature aswell is adding consumption via mqtt or something similar. Im using evcc for my ev and it would be really nice if the scheduled charges for the next day are shared with victron.