Peak shaving with 6 Multiplus 2 not working?

[Kolja]

Hi, I’ve recently updated my setup from 3x Multiplus to 6x Multiplus, (as mentioned here ESS Speed of 6x Multiplus II way slower compared to 3x Multiplus II )

On top of the ESS regulation I now also see a strange behavior for peak shaving.

My understanding (from https://www.victronenergy.com/blog/2024/03/19/venus-os-v3-30-peak-shaving-for-ess/) of Peak shaving was that the ESS would see EVERY load behind my meter (not Multiplus (this is what is limted with the Grid current limit)) and adjust charging power to meet the defined “maximum import/export current per phase”.

I understood “There are two new AC current-limit settings, and those limits will apply to the complete system. Therefore they now include any high-power loads which are usually installed in parallel to the inverter/charger system such as EV Chargers and Heat pumps.” in the way that loads can also be on ACin1 and I also though my ‘old’ system did exactly that. ChatGPT is quite certain that only loads in ACout will be considered for peak shaving.

Can someone clearify this already ? (although this isn’t really my problem )

For my testing today I moved all loads to AC-OUT, so peakshaving is somewhat similar to Gridlimit (all all current has to go from AC-IN1 to AC-OUT)

Why I need peak shaving

I have a 50A (~34kW) main breaker, can charge my batteries with 24kw (6x4000W) and have 2 EVs which can charge with 11kW and 7.7kW and a heatpump that usually runs around 2kW + house loads. So I can easy kill my breaker.

My settings

Peak shaving is enabled (and can’t be disabled for some odd reason (but I’ve read multiple times that is ‘normal’ right?) and for this test set to 25A

25A => ~17kW

for this test I’ve also set the grid limit to 32A => ~22kW

Switching to “keep batteries charged” leads me to ending up at

having L1 already go over the 25A

So: 32A Grid | 25A PS = 28/19/29

activating one car charger (car + heatpump + loads → 18kW)

now we’re running already at 32A or ever slightly higher

So: 32A Grid | 25A PS = 28/19/29 → 32/18/33

when I reduce the grid limit to 25 doesn’t effect the 32A

reducing the peak shaving to 20A

finally some effect .. we’re reducing to 27ish A

So:

32A Grid | 25A PS = 28/19/29 → 32/18/33
25A Grid | 25A PS = 32/18/33
25A Grid | 20A PS = 25/15/27

I thought this was running spot on with my old 3x MP2 setup .. why is this so off with the 6 MP?
Am I doing something crazy wrong? Any Idea? I’m running the latest stable firmware and VenusOS and the ESS assistant is up2date

Any help / idea is welcome, Thanks

Kolja

PS: also fancy

30A are sometimes 5.5kW and sometimes 8.5kW
(I’ve double checked with dbus-spy and the Wattage is correct .. not sure where the Amps are coming from)

[Kolja]

After countless testing and reconfiguration I found the issue:

PowerAssist was off on one of the six Multiplus.

After enabling it Peak Shaving is working again .. and the switch is now also available (not longer grayed out) … easy fix

[lxonline (LX)]

@Kolja

PowerAssist was off on one of the six Multiplus.

Usually this is disabled by the ess assistant. So that is interesting. Are you sure it is loaded correctly on all devices in the system?

@Kolja

ChatGPT is quite certain that only loads in ACout will be considered for peak shaving.

Mentioned in the manual and in the blog post. It works on both sides if there is an energy meter.

[Kolja]

What I’ve seen on com.victronenergy.vebus.ttyS2 was

Devices/[0..5]/Settings/PowerAssistEnabled

being “1” for 5 out of the 6 devices. After downloading the config, finding the one that didn’t have in enabled (check in ve.configure) and reuploading the “Peaks shaving” button was not longer gray and I can now turn it on and off ..

Following up on that I started to charge my battery with 16kW which was exactly the allowed Ampere on the gridmeter. when my turned on my ev charger the current for the battery went down to again respect the limit.

There is however one uncertainty still remaining:

Untill yesterday all my load were connected to ACout, but the EV charger. Yesterday evening I moved the EV charger to ACout to test if chatgpt was right (which I was 98% certain it wasn’t) .. turns out that peak shaving didn’t work with all loads on ACout, so I moved the EV charger back to ACin.

Only after that I found the missing power assist, but to be able to update the system I had to switch ALL loads to ACin. I have a maintenance switch for that, but it is still a pain as all devices obviously loose power and smart homes take some time to be up and running again…. So I’m trying to do this not too often

So ultimately the entire house right now is still on ACin, this is the addition difference, besides having PowerAssist enabled. I don’t thing that having all loads on ACin enabled the switch, which is why I’d suspect the PowerAssist bit .. but I’ll gladly test again after moving all loads back to ACout. That will have to wait as I’d like to also fix the ESS issue from the other thread and that might require reconfiguration

[lxonline (LX)]

Ok. So you also have mods and other interfaces with the system. Useful to know.

On the Gx itself -confirm you have the system set up to know that the system has loads on the ac input and output?

@Kolja

Switching to “keep batteries charged” leads me to ending up at

This by the way usually switches off/dials back quite a few ess features. So having it on optimised with batteries at 100% is a better option here.

[Kolja]

What do you mean by mods?

and yes the system is configured to have loads on both ends.

@lxonline

So having it on optimised with batteries at 100% is a better option here.

good to know, thanks

[lxonline (LX)]

@Kolja

com.victronenergy.vebus.ttyS2 was

Devices/[0..5]/Settings/PowerAssistEnabled

You are digging around… Usually there is something interfacing …

[Kolja]

I quite like dbus-spy .. great tool to see what effects the changes in the GUI have

[Kolja]

so peak shaving is set to 42A per phase .. turning on the EV charger leads to:

yep, ESS regulation is still bad, grid feed-in isn’t 0

setting min soc to 100% leads to

which is pretty spot on 42A

so I’d say “it works now”

next test is switching load to ACout, as soon as the ESS loop is working better (or I give up)