Peakshaving ignores external meter, loads on ac-input not taken into account

Beta testing Venus OS beta testing
, ,
21

Quite ironically, yes, good thinking, but unfortunately we already found that one and fixed it in Venus 3.70. A very high setpoint could cause peak shaving to be overruled.

This also happens during charging. When we charge, and that setting that feeds excess DC PV into the grid is enabled, it is done by internally setting the setpoint to a very large number. The calculation I referred to earlier replaces this “large number” with a calculated number that is low enough to pass under the bar. And this calculation is done periodically (every 2.5 seconds) to react to other loads that may come and go.

It also takes into account power factor. Because we have to calculate a power value that will result in the correct amount of current. To calculate power factor, it constantly works out the ratio between VI (voltage times current) and P (watts), that is, it estimates cos(phi). It does it this way, instead of using the power factor from the meter, because not all meters have power factor.

At your site, when you charge, the power factor is very close to 1. I actually spent quite a bit of time trying to work out if this might be it. The symptoms would be explained if the power factor was much lower than estimated. But both the Multi, and an EV charger, is typically very good on power factor. So I don’t think it is this.

There is one possibility, which would be if you have the “Weak AC” option enabled in your VE.Configure settings. Then charging with the Multi has a lower power factor.

That’s another thing I considered, and this is not ruled out yet, but that would imply that the meter is giving me wrong current readings, when those current readings are the very thing we use to say that there is a bug. So doesn’t seem likely.

If you have this set to yes, then the code bails out (skips any control of PV inverters and the Multi) in exactly the same place as it did in 3.67… so also unlikely.

But I do like the theories, which is why I am keeping the conversation here.

I will take a look at last night’s data again, thanks for your patience on this.

1 Like
22

Hi Izak,

I am pretty confident that I did not change Weak AC option from the default in VE.Configure. I cannot check unfortunately, because of the battery upgrade. My system will be off line for one or two days.

For now I ran out of theories :frowning:

Rob

23

Hi Rob,

I just found a bug in this thread: ESS not working correctly in V3.70 when peakshaving - #31 by iburger

And the line of code involved is identical, although in a different branch. It’s the calculation of how much battery power to use to get the grid to the right value. We’re making progress.

24

Hi Izak, I have the same problem. Only noticed it now. Has been working for me like forever. But today the consumption was over 26A regularly. I have even seen 29A+. Using DVCC I limited the charging current to survive the afternoon. <290452>

Would you mind sharing the line of code or change it for me ?

25

The issue has been found. The output loads are being accounted for twice, in the specific case when we’re charging with a import current limit defined.

Unfortunately this is not open code, but I have already applied a fix to your system. And when @robmeerwijk is done with his battery maintenance we will do the same to confirm the fix.

This will go into Venus 3.71.

Thanks to all involved!

3 Likes
26

Set “keep batteries charged” and removed the DVCC limit. All phases, consumption un-evenly loaded, where at 25A for a long time without issues. So your fix works fine for my setup/config.

2 Likes
27

Hi Izak, battery upgrade done. System is back on line and on grid!

28

I just installed beta 3.80~4. I read in the release notes that it should fix peak shaving issues in 3.70. I will test next time when my car needs charging again. I will have to decrease peak shaving limit because of PV-output :sun:

29

@iburger Hi Izak, I have tested peak shaving with 10A limit. Working like a breeze now. Thanks a lot. I have reported a new issue in the meantime with DESS. Maybe you have seen it already.