Running Beta v3.70~61, MP2 48/10000 with 32 kWh battery, grid meter and SolarEdge AC coupled solar (not doing much at the moment)
I am on Octopus Cosy, as I have a heat pump, which has three ‘cheap’ periods a day, plus a 3 hour peak period, these and the prices are loaded into DESS.
What I have noticed is that the batteries are not being charged throughout the low cost periods, often charging in only part of the period, which means the MP2 is running at full blast - which is noisy, especially as one is an the early morning period.
Why can the DESS not schedule the charging to be even throughout the lower cost half-hours?
the scheduler generally does charging from Grid “as late as possible”. This is the way of minimal forecast error being introduced and - in your case - optimizing solar2battery by not eventually overcharging in the morning. But you can force it to utilize multiple windows by configuring a lower battery charge rate for your system than it could actually do.
i.e. if you pretend your maximum charge rate will be 4000, it’ll only schedule for a 4kW chargerate and therefore has to spread across multiple of your slots to achieve the desired amount of energy being charged from grid.
You can also do that assymmetric, say 4kW Charge / 12 kW discharge.
Many thanks for explaining how it works, I can see the logic when there is significant solar, but at this time of the year I struggle, not helped by my neighbours tall trees shading when the sun is low, so I have to use grid power, and I aim to only import during the three low cost periods, and minimise grid imports the rest of the day.
I will try your suggestion, as I would prefer a longer, but quieter, charging window, especially for the late night and early morning slots.
That should work. I also limited mine to 8kW cause I don’t want things to run at 100% to save some cents just to come down to the very cheapest windows.
But since in your case all are equally cheap, there is not even a financial drawback.
@dognose I have noticed another strange DESS behaviour today, it is 19:00 and the tariff cost are Standard £0.24 KwH, so I would like to avoid paying this, but the system is not using the battery but taking power from the grid, despite the battery being 50% SOC and the minimum set to 30%, and a new charging window available in 3 hours, by which the forecast SOC is just above the minimum.
I would expect DESS to attempt to wait to see if we got close to the minimum, say at 21:30 before deciding to import from the grid, as we may not need to import before the cheap period starting at 22:00.
I’m not part of the scheduling team, but to explain single situations you need to always look at the whole whole schedule for the upcoming 48 hours as sometimes even “tomorrow morning things” will impact what the scheduler does today.
Mind the scheduler is planning based on forecasted solar and forecasted consumption. So the plan is always “mathematically ideal” IF the forecasted values are exactly happening as forecasted. Deviations in solar / consumption may sometimes make past actions turning into “unideal behaviour” when looking back from a future point onto the actual values.
For example when it schedules for idling during moderate price hours, because it expects huge consumption in the most expensive hours: If it happens like that, all good, money saved.
But that day, you didn’t turn on the dishwasher, so there was less energy required than forecasted - and when looking BACK on the schedule you see “Idling during moderate price hours and beeing 10% above minsoc in the morning - not ideal”.
I am trying to understand the logic behind DESS, as I do like the idea behind DESS, especially when we have more solar than during winter, as I should be nearly self-sufficient in summer and I expect it will be a great solution then, but in winter it is struggling.
I do appreciate that sudden, unplanned consumption can cause the actual values to differ from the forecasted, resulting in the need to import at less favourable prices, but looking at today’s forecast I see again a possible import outside my cheap periods, although tomorrow looks better.
However, DESS is forecasting charging my battery to only 65% this afternoon, and it is projecting a completely wrong solar forecast than the UK Met Office (no sun but heavy rain and gales), I would expect DESS to be charging the battery to say 80% to provide a buffer against an increase in consumption or solar being less, than forecast.
I would like to retain at least 30% to cover any supply disruption, so that should be a absolute minimum, at which point the system will import, ignoring any tariffs set, but under normal use we should be able to set a buffer of say 10%, so DESS works between the 40% and an upper value of say 80% normally.
Perhaps something for the scheduling team to consider, as I have just done a search and found other similar suggestions being made.
There are times when the ‘I’ in AI is not that intelligent, but I would prefer to use your solution than start to code my own, life is too short.
I think the scheduler calculates the amount of energy “that would be missing based on expectation” for the active period and only charges that from grid.
It’s not doing some pro-active “Oh, it’s cheap, grab another 20% now for whenever it’s required”.
The reasoning behind this is simple: That would have way higher potential to end up unideal. If it charges up to 80%, then consumption is lower than expected and solar WAY higher - you’ll hit 100% soc and have to feedin for almost no earnings.
And that is the financial worst-case:
If you then have to feed in 10 kWh for 8 cts because it bought 10 kWh over expectation for 24ct - it’s an actual loss of money. The error by not stockpiling energy, so “running out of energy because it bought to less” - is not a loss, it is just a “missed saving opportunity” so from a financial impact view a less severe situation, because there are STILL savings compared to a regular ESS-Operation - just not as much as there could have been.
That would make perfect sense if, and it is a big if, solar was making up a sizeable portion of the energy, at the moment being winter even the forecast is only expecting the solar produce 2-3 kWh today - very optimistic, and all of that would be self-consumed.
I would be happy to charge up to 80%, leaving 20% (6.5 kWh) spare to cover higher than forecast solar surplus - over my self-consumption, which would enable me to avoid importing during the Standard 24p and 44p periods while leaving me 30% capacity for any supply failure.
The algorithms need to adjust for the shorter days and lower solar capacity in winter, against summer when solar is a significant element of any supply here in the northern hemisphere.
