I noticed a few days now that the charging of the battery is not scheduled during the cheapest hours but a few hours later when the price is already increasing.
I can’t seem to find a logical explanation. Recently the bulk of the solar was changed from AC connected to DC connected with a victron MPTT, so in theory the MP2’s should be able to have the capacity to take the extra charge from the grid during the cheapest moments.
That is just a display issue of the VRM graph. (As the current hour passes by, the forecasted values will go from 4/4 over 3/4 down to 1/4, as depending on “now” 1,2 or 3 15 minute-blocks drop out of beeing a “forecast”)
Thanx for the feedback. I’m running in trade mode. Meanwhile I resetted my solar statistics, but this didn’t help (looks like no difference is made between AC & DC solar)
I also noticed, I think a different bug than what you mention @UpCycleElectric :
The total charged per hour is reflecting the DESS settings:
DESS power settings are notoriously misunderstood and sparcely documented. Here the basics:
DESS settings are only used as input parameters to the VRM scheduling algorithm. They are completely separate from any other (power) control parameters and actual physical limitations of the system.
For the DESS algorithm to function correctly it needs to know:
Battery capacity
Roundtrip efficiency
Maximum AC to grid power (I set 4.4kW for a Multiplus II 5000)
Maximum AC from grid power (I set 8.5kW to reflect 230Vac 35A mains fuse, I also use secondary charger to reach that power)
Maximum DC to battery power (I set to 9kW, plenty high for the maximum 8.5kW AC charge power-> 7,8kW DC charge power)
Maximum DC from battery power (set to 50kW to enable high imaginary EV loads at the end of the known prices period, in practice this never happens because I shift that imaginary load forward 24 hours every afternoon, and there is no load anyway, this just to make the scheduler starting charging in time every morning, thinking there will be a large load right before midnight)
MinimumSoC % (I set 20%)
These parameters allow the algorithm to plan to buy at 8.5kW and sell at 4.4kW while leaving heaps of room for other DC loads/sources.
Unfortunately this is not resulting in an optimum schedule, see (all of) my post(s) about the design flaw based on the assumption that either or both conditions apply a) solar production is large in relation to battery capacity and b) inverter charge/discharge power is large in relation to battery capacity. As soon as your battery capacity is more then about 4 x inverter power (* 1 hour), as in you can charge/discharge longer then 4 hours at full power, and/or when your solar is small, and/or when market prices are very low/negative. The algorithm will always fail to start charging in the morning, because it cannot see enough sell opportunity until midnight same day. Only after the new prices roll in, it can see plenty sell opportunity beyond midnight into the next day, but then it already missed the buying opportunity.
This flaw affects every DESS trade system, it’s negative impact is directly related to battery capacity and inversely related to solar capacity. Setting an imaginary large EV consumption right before midnight, and shifting that forward every afternoon, fully compensates this flaw, but is still an ugly hack, that Victron continues to ignore the root cause off.
From the looks of it this is because you use the DC battery power settings to tell the scheduler not to exceed 5kW total DC charge power.
You could increase that power all the way up to your actual battery charge (discharge) limits and set the AC (to/from) grid power setting all the way up to your AC mains fuse limits, or lower to your inverter / charger actually achievable AC performance, or in case of AC coupled solar somewhere in between.
Actually, to make things worse, you happend to stumble upon an actually bug here, but different than you might suspect, this is a documentation bug.
In your screenshot you show your 5kW battery DC power limits (as is evident from the fact it is set in the battery power limits section). The bug is in the description text that you underlined in red, that should read battery not grid. Apparently the description of the grid section has been copied there verbatim, implying wrongly the battery section also concerns AC grid limits while it concerns DC battery limits. That’s a pretty devilish documentation bug in this already complicated matter..
See here to spot that both sections have the same description:
Definitely some strange things with the settings. I also tried cranking that setting (Battery max) up but and that works when the sun is out. But when there is no or little sun that settings is forecasting a lot of power coming from the grid which is not possible due to the multiplus limits.
So either there should be an extra set of settings for the charge power from the MP or the algorithm should take the MP size into account (implicitly)
I rather not turn the grid settings down, as this will have new side effects with big loads.
No it won’t. These settings do nothing with any actual power control behaviour. They only influence the DESS scheduling algorithm, which output is primarily the planned SoC schedule (with matching Strategy and Limitations flags). For all practical purposes the power control loop only attempts to follow the planned SoC schedule, but whether it succeeds or fails still depends completely on the actual system characteristics and settings that are, again, completely unrelated to the DESS scheduler power settings.
My approach would be to first get your AC grid power settings tuned in correctly to allow DESS plan for full power charge / discharge (during the obvious highest / lowest price windows. To achieve that, the DESS DC battery power limits can be set very high, just to exclude DESS from attempting to DC throttle.
It’s been a recurring theme that the documentation on the DESS settings is confusing at best, still you will have to make sure you understand exactly what they do before attempting further tuning. Hence: eliminate the influence of the DESS DC battery power settings at first (set high) , figure out what the DESS AC grid/system settings do first. And directly thereafter, make sure you know and set your real battery capacity and a realistic roundtrip efficiency. Unfortunately for the latter you probably need to learn a little about using Node-RED because that setting can’t be set in the GUIv2 as far as I am aware.
Any way, the first objective is to get the DESS SoC planning to be spot on during those high power buy/sell windows (and any deviations clearly relatable to solar/consumption deviations from forecasted)
Although I am personally not aware of any better documentation or instructions on this DESS settings topic, I still advise doing a search for it as well, official documentation, forum discussions or even external blogs/YouTube, it’s such a common hurdle to get right chances are somebody documented it much better than I can explain in a couple of posts.
That exactly why I set my DESS AC grid (it’s called system in the settings page) to the 4.4kW my MP-II 5000 can deliver to AC grid and to the 8.5kW that my chargers (MP-II plus Huawei PSUs) can take from AC grid (without blowing the 35A mains fuse) resulting in about 7.8kW DC charge power. See screenshot above.
PS without those Huawei PSUs I would lower the AC charge power to whatever the MP-II draws at full charge power, around 52V x 70A x 1.1 gives 4kW AC, 3.64kW DC, the difference heatloss.
My only blindspot here is how to take AC coupled solar into account here, maybe somebody else can jump in on that one.
I am sure they do, whether they will include work on DESS is another question though, best chance on success is by taking the lead yourself and report what you learn here
