Vaillant: add reboost and hysteresis

[andig]

#30636 added re-boosting for OHPCF. We should do the same for API chargers. It remains open to see what should be done about SG Ready (see #29171).

[Johannes-del]

Data point from a live aroTHERM plus (myVaillant/sensoNET, DHW in mode: pv, limitSoc target 60 °C).

First, the vaillant charger already reboosts today: set(Boost) fires StartHotWaterBoost and a goroutine re-issues it every 15 min until the loadpoint disables (charger/vaillant.go, same in 0.310.1 and master). So in practice the pain isn't a missing reboost, it's the missing reboost hysteresis (the open #29171):

• With continuous midday PV surplus and limitSoc at the DHW target, evcc re-enables the boost the moment the tank drops a fraction below target. The result is constant nibbling at the ceiling: the heat pump short-cycles and the circulation pump runs more or less continuously, instead of one clean charge.
• A reboost deadband fixes it: only reheat once DHW has fallen a real band below target (e.g. reheat at 50, ceiling 60), which gives a single healthy ~35 min cycle per drop instead of ping-ponging at 60.
• Polling subtlety for whoever implements it: evcc stops exactly at the target and the tank falls just below within the same poll window, so a naive "temp >= target => block" edge is easily missed. We additionally latch to "block" on boost-ended (charging = false) with temp near the ceiling, which catches it reliably on the first sample after the boost ends.

We currently emulate this externally by toggling limitSoc between 50 and 60 based on the DHW temperature (evcc exposes it as the loadpoint's vehicleSoc). Happy to test a native reboost/hysteresis on real hardware.

[CiNcH83]

the heat pump short-cycles

I have never seen this. The heat pump compressor is only restarted on Boost if it goes below the "Hysteresis Speicherladung". For me, only the circulation pump has been a problem.

There is only one circumstance under which it indeed restarts the compressor and that is if it hasn't reached its DHW target temperature (Solltemperatur) before. But if you set that lower, you will not reach your goel of a dynamic target temperature? Your problem is that the system hasn't been built for what you are trying to achieve. That said, OHPCF will probably be your friend, where you can set a different target temperature for boost, see #30636.

[andig]

@Johannes-del could you try PR? I've added hysterisis for reboost, though not for loadpoint. If loadpoint switches enable it will always boost, independently of hysteresis.

[Johannes-del]

@andig thanks. That's exactly the deadband, will test it on the aroTHERM. Probably tomorrow.

@CiNcH83 fair correction. You're right the compressor has its own "Hysterese Speicherladung" and doesn't hard short-cycle. On our unit it's ~4 compressor starts in 3.8 days, which is fine.

On the PR: as @andig notes, the hysteresis gates the reboost loop but not the loadpoint enable. Our nibbling is mostly at the loadpoint level. evcc stops exactly at limitSoc, the tank drops just below, the loadpoint re-enables and boosts again (which is why our workaround toggles limitSoc rather than spacing reboosts). So the reboost deadband should help the "held enabled continuously" case, but probably not the enable-flap at target. I'll test both and report with numbers.

[andig]

So the reboost deadband should help the "held enabled continuously" case, but probably not the enable-flap at target.

Fair point, let me fix that.

[andig]

@naltatis added an easy fix for hysteresis. This is no general pattern though. We might need to move this to the loadpoint. However, devices would still need to know the value for skipping re-boosting. For time being it seems fair to leave this on the device level rather than loadpoint?