Repeatedly charging and discharging the battery

[gcoan]

Describe the bug Noticed this morning that predbat is letting the battery solar charge then force discharge, then repeat. This can't be efficient

7.16.16 and secondary order slots OFF

Thought it must be 7.16.16 causing it so reverted to 7.16.15 with secondary order slots OFF, much the same:

Tried turning secondary order slots ON with 7.16.15 and its still similar

Expected behavior Either hold the battery level low or let it charge and stay full to the peak would involve less battery conversion losses

Previous plans with secondary order slots off have been fine

[springfall2008]

This is odd, can you try the updates on main also and see if it makes any difference?

[gcoan]

I had put it back to 7.16.16 and still trying the charge and discharging

Eventually with a load of force idles I managed to stop it doing this

Will try main now

[SwiftRR]

I am on 7.16.16 with secondary order slots ON This is what I see, running on until sun sets.

Rob

[gcoan]

Looks like @SwiftRR is having the same behaviour

I tried main and cancelled the force idles , same thing

Putting the force idle back in, revised plan has less battery cycling but projected that it's slightly less profitable- which makes no sense, charging and discharging the battery will incur more conversion losses than just exporting straight from solar. But might explain why predbat selected this plan if the loss calculation is wrong?

Edit: noticed that the solar for the current slot has changed by nearly a kw in two runs a few minutes apart ?

[springfall2008]

Do you have the conversion losses set correctly?

[bensebborn]

I have lots of solar export today and mine is correctly holding charge and exporting the solar directly so doesn’t seem to be affecting me

[SwiftRR]

I have an GE AC coupled inverter. Hybrid inverter switch is OFF Conversion losses are 4/4/4

On today’s snapshots, daytime charging of battery mainly by solar with ’popup from grid’.

I am happy to tweak settings to help test the issue.

Rob

Rob Ritchie @.***

On 20 Apr 2024, at 14:29, Trefor Southwell @.***> wrote:

Do you have the conversion losses set correctly?

— Reply to this email directly, view it on GitHub https://github.com/springfall2008/batpred/issues/975#issuecomment-2067674528, or unsubscribe https://github.com/notifications/unsubscribe-auth/AQNA33DYRVMYLNGMICSL443Y6JUUTAVCNFSM6AAAAABGQJT2JCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDANRXGY3TINJSHA. You are receiving this because you were mentioned.

[gcoan]

Do you have the conversion losses set correctly?

My loss figures have been the same for ages, just checked and they are: inverter 4% battery charge 5.5% battery discharge 5%

hybrid=true

battery metric cycle cost=0

metric min improvement=0

metric min improvement discharge=0.2

I don't think I have seen this behaviour from predbat before.

Tomorrow is doing similar, not quite as extreme but still planning repeated charge then discharge when a hold charge and export the restwould normally be more efficient:

Can provide the logfile if you would like to see it

[SwiftRR]

For comparison with Geoffrey's times for Sunday, this is my predbat plan:

Tariff Agile import, 15p fixed export.

I am in Yorkshire region and my import rates are about 0.6p lower than Geoffrey's.

Conversion losses 4/4/4

hybrid OFF

Secondary order slots=ON

metric battery cycle cost=0

metric min improvement=0

metric min improvement discharge=0.1

Nordpool ON

Rob

[gcoan]

@springfall2008 think this is probably related to #924 and #861 both of which are about the behaviour of input_number.predbat_min_improvement_discharge that doesn't seem to be working properly

If I progressively increase the min_improvement_discharge I eventually get to a plan that "looks right" and isn't charging and discharging the battery to make a profit

0.1:

0.2:

0.5:

1:

3:

6:

[gcoan]

Had another look at it this morning. Using the version on b2663b8 on main

Secondary order slots off

Initially metric_min_discharge_improvement was 6p and the plan looked OK:

changed back to 0.3p and a series of discharges were introduced:

and 1p, slightly less discharging but similar:

Looking at one of these in a bit more detail, discharge and export is at an effective rate of 13.68p vs just keeping the charge at 100% and exporting the solar. The first (discharge and solar recharge) incurs the battery discharge loss (5%), the inverter loss (4%) and on recharge the battery charge loss (5.5%) and a second set of inverter loss (4%). Compare to just exporting solar straight away which will only incur the 4% inverter loss and should be more profitable.

Increase metric_min_discharge_improvement to 5p and then 6p:

In both of these predbat appears to be discharging and exporting at 13.68p effective rate to then import and charge at 15.77p effective rate?

Increasing metric_min_discharge_improvement to 8p stops the discharge and recharge which is the correct plan but is having to artificially constrain using the battery. I would rather be using the battery if it were truly profitable to do so

[SwiftRR]

Interesting. For comparison, here's mine with metric_min_discharge_improvement at default of 0.1 p !!!!!! calculate discharge on charge slots TRUE combine charge slots TRUE combine discharge slots FALSE calculate secondary order slots TRUE ... and running with losses, 4/4/4 on hybrid inverter FALSE Looks much steadier.

Rob

[gcoan]

metric_min_discharge_improvement at default of 0.1 p !!!!!! calculate discharge on charge slots TRUE combine charge slots TRUE combine discharge slots FALSE calculate secondary order slots TRUE ... and running with losses, 4/4/4 on hybrid inverter FALSE

And my settings are:

metric_min_discharge_improvement at 8p to stop the discharging calculate discharge on charge slots FALSE combine charge slots FALSE combine discharge slots FALSE calculate secondary order slots FALSE Losses, 5.5/5/4 on hybrid inverter TRUE

Also Calculate full second pass (slower), Calculate plan faster (less accurate), Calculate tweak second pass all FALSE Metric battery cycle cost 0p

[SwiftRR]

I also have: Calculate full second pass (slower), Calculate plan faster (less accurate), Calculate tweak second pass all FALSE Metric battery cycle cost 0p

My plan seems to be leaving itself. This one has rumbled on for a while now. Is it 'Over to Trefor time'?

Rob

[gcoan]

I think it is one for @springfall2008

There's the other two issues also about min discharge improvement not seemingly working "quite right"

[springfall2008]

Closing this old ticket, please open a new one if you have an up to date question