question

Regarding 1

Yes and no. The MPPT will try to grab and hold onto the float state once it has been reached. It can dial the output amps up to the rated amps it can deliver.

However, since Watt is the mathematical product of Amps and Voltage and float voltage is lower than absorption voltage, Watts (needed to heat the water) produced are lower in float than in absorption state.

No opinion on 2

To 3

If your batteries are charged by 1100h and the Mppt goes into float your 2000W DC Heater (a lot) is to big for the current maximum possible yield of say 1800W (very likely only ever midday on a good day ie barely ever) it will drop out of float within seconds, stay in absorption for a while and go into float and again drop out within seconds. No actual water heating going on then. 2000W x 0,1h of runtime in a not blue sky day makes 0,2 kWh of not ice cold water.

If you had a 500 W element on the same day it might run from 1100h till 1600h right through.

500W x 5h makes 2,5kWh or one washing machine worth of at least warm water.

Now for the PWM device on the same day, it can variate the power going to the 2000W heater from 0 to 100% depending on availability.

It switches on at 1100h and goes up to 1800W, holds it up there till 1330h, goes ever down to 500W around 1600h and keeps going while still dialing down till the sun sets.

You be looking at say 6 kWh of hot water. More than ever and always as much as possible. The smaller heating rod can be a KISS alternative if chosen sensibly. It is a compromise of running time and output that need to meet your available power calculated as above. Aim low for bad weather.

To 4

Your choice. I change it by the season. In summer the batteries will be charged fully sometime during the day anyways. Although a PWM device will burn all available power so there is nothing left to pamper (lead) batteries or needs to be adjusted to leave enough excess. About a month ago I changed back to waiting for the batteries to fill first. Afterall it is called opportunity load.

Sadly Victron has nothing proper in that regard on offer although they can provide the necessary measured data already and could offer a solution fairly easily. Until then its tinker time or going with the competition.

If you have a larger MPPT model with relay and a BMV, my recommendation for a BASIC dump load control would be to use both relays wired in series so that they both must be closed for the dump load to be activated.

The most important part with a non-modulated dump load would be to choose an element with a suitable/moderate power rating. If you go too high the 'spare' solar will simply be insufficient (most of the time) to cover the load and depending on your conditions, it will then be quickly deactivated.

I would propose to use SOC% and low battery voltage on the BMV with the relay output inverted (resulting in the opposite behavior & AND logic to activate the dump load).

It would be key to set the voltage 'clear value' just below the charge voltage in float phase and the voltage 'set value' just above the battery resting voltage.

That way you can be fairly confident that the dump load will only be active when the solar can support it, with very little energy ever taken out of the battery.

I would also propose to trigger the MPPT relay based on float phase being achieved (per your thoughts), that will prevent the dump load from effecting the bulk and absorption phase (otherwise the absorption voltage might not be reached), as well as prevent the dump load from being activated early morning / late evening when the MPPT is in bulk phase with little solar available + at night when the MPPT is off (if the BMV conditions were both met somehow, like with an external charger).

In terms of how the MPPT works, it just tries to maintain the charge voltage setpoint for whatever charge phase it's in and it can deliver the full rated charge current in any charge phase.

It is true that the charge power (W) would be slightly higher at a higher voltage, but this should not be a concern since the fixed power dump load shold be sized much smaller than max MPPT output anyway.

With a basic control logic like this you can't expect to extract every last % of available solar, but it's better than nothing and will 'work'. So ensure that all choices are a good trade off to maximize the overall energy provided to the dump load in a wide range of solar conditions.