A well-known problem is that irregular high consumption, such as charging an electric car, “destroys” the consumption forecast.
What is the current status in this regard, are there already concrete approaches to fix this or when can these be expected?
As we will soon get our first electric car, the topic is becoming interesting again.
Hi,
I am not aware of any upcoming changes there, neither did I note any Issues with the forecasting on my EV.
Sure, when switching habbits, it takes some time for the scheduler to adapt, but that applies for any load(s).
But if you have an example what you mean by “destroys the forecast”, I could forward that to the forecasting team to have a look at.
My consumption in summer is always around 20 kWh/day, which the forecast always showed well.
Last weekend, a friend came to visit with his EV, which we charged. See extraordinary consumption 15 & 16.08.
The following forecasts are therefore clearly too high, as no EV has been charged for a long time (months). Very high consumption is expected again for 22 & 23/08 in particular.
That’s what I mean. The two charges on 15 & 16/08 were exceptional outliers which should not affect the future forecast.
There is currently no way to classify outliers as such so that they do not affect future forecasts.
Somehow this needs to be better accounted for with EV charging. EV charging often takes place at irregular times with exceptionally high consumption peaks. It should be possible to take this into account or influence it better somehow.
@dognose Phrased another way, it would be nice if certain measured loads could be specifically excluded from forecasting. I irregularly charge my EV in the evenings on random days, so the system projects a charge every day roughly due to the compounding effect of the irregular charges. I’d rather the system forecast ignore those potential EV charges.
I think beeing able to completly exclude the EV from forecasting would make sence for many users and situations.
In fact, that is an issue for me as well, I forgot about during summer 
: In Winter I usually charge the EV at night (cheap wind from grid) and therefore do not drain the houesholds battery. Hence, DESS shouldn’t mind this consumption as well in the scheduling, as there is no battery beeing used for it.
To be honest… a year ago the consumption forecast was configured completely differently. Yes, back then one car charge would lead to extreme fluctuations. But since a change in November 24, the three BEVs are lost in the background noise for me. Yes… with an unscheduled electric car charge the forecast goes up, but after 2-3 days it has been put into perspective and after a week it is no longer visible in the forecast. I want to name 2-3 extremely positive examples. First situation: our tenant bought a BEV and connected to our grid… the system integrated it within a few days… perfect. Then our daughter moved out and stopped charging her BEV regularly. The system understood this within a few days too. Now she often comes over on weekends and charges naturally; the system is already waiting for her. I was most positively surprised by the system at the beginning of August. A few days of bad weather… the BEVs were not charged for a few days… the system coped well. I don’t really know why, but I bought it at the last minute, as I couldn’t have made a better decision myself. At the moment, I personally think it’s good, but experience has shown that that can change quickly in a different situation.
I don’t believe a VM3 can be tagged as an EV charger, but that would make this somewhat easier (given it can communicate via Ethernet)
Then somehow set VRM to ignore certain metered loads from the consumption forecast.
Well, seeing the EV just an example. A more generic approach ofc. would be preferable. I can imagine (I do not know) that there are different consumption sources beeing taken into the forecast, so, eventually that list could somehow be exposed, allowing to enable / disable certain sources with a checkbox or something.
I will propose this to be looked at and what options there are.
That would be amazing. I’ve seen a handful of threads with folks asking for similar ability.
I’ve seen another vendor do this a bit more manually (maybe Sige Energy?) - where you go into their app and manually tell the system to “ignore” a certain spike in load, excluding it from the forecast.
Much rather prefer the automated approach though, by selecting metered sources.
Another thought I had on this topic… for larger systems (I have 4x 10k Quattros), the idle power draw is ~400-500W. This draw isn’t included in the consumption forecast, though can cause issues when nearing min SOC (i.e., the system can hit min SOC too soon during a peak price period). It would be nice if there was a way to account for this consistent “drag” on the DESS plan.
There’s another thread where mpvader was going to look into whether it’s as possible to permanently disable PowerAssist in v3.70, which would reduce this drag to some degree (since it lowers the idle power consumption of the system). Bug?: Peak shaving drains battery permanently despite minimum SOC - #16 by JeroenSt
