Unreliable Weather Forecast, Solar Behavior, DESS Timing

Unreliable Weather Forecast Causing Faulty DESS Predictions

After resetting the weather forecast data in Venus OS, the system initially provides a realistic forecast for about two days (e.g. around 20–25 kWh/day). However, after that period, it suddenly jumps without any clear reason to highly unrealistic values such as 90 kWh/day or more.

This has the following effects:

Dynamic ESS receives incorrect input, which leads to faulty assumptions about available energy.
Battery control decisions are then based on these unrealistic forecasts, resulting in premature or delayed (dis)charging.
The “Energy to be Expected” value inside DESS appears to partially track the forecast, making the behavior even more inconsistent.
These forecast values regularly exceed the physical system capacity. We are grid-connected with 3×25A, and actual production should never realistically exceed ~17.5 kWh/day.

This is not only confusing but leads to concrete inefficiencies and DESS behavior that is not aligned with system capability or grid constraints.

  1. Solar Behavior at Maximum SOC

When the battery reaches its maximum state of charge (SOC), the system correctly stops charging and prioritizes self-consumption — particularly when dynamic prices go negative. However, even during extended negative pricing periods, solar production is not being curtailed or turned off, resulting in unnecessary and uneconomical energy export to the grid.

We would expect Fronius and/or the MPPT to be throttled or disabled in such cases, but that is currently not happening.

  1. DESS Timing and Minimum SOC Not Respected

Dynamic ESS appears to react too late to pricing and weather signals. This causes the battery to remain at too high SOC levels during the early part of the day, blocking solar production later on.

For example:

Today’s minimum SOC was configured at 15%, but the system only discharged to around 40%.
This left insufficient buffer for solar generation, and the system had to limit or stop solar input during high production hours due to a full battery.

Request for Support

Could you please advise:

Is the weather forecast issue a known problem in Venus OS, and is there any way to force accurate or capped forecasts?
Why are Fronius and SmartSolar not being curtailed when prices are negative and SOC is full — is this a configuration issue or a limitation in current DESS logic?

We would appreciate any insight or workaround, as these issues significantly impact system efficiency and make automated control less reliable.

407704

DESS Logic Struggles with Panel Orientation and Oversizing

We have also observed that the current DESS algorithm appears to struggle with more complex solar layouts, especially in systems where:

Panels are oriented in both east and south directions, leading to a more extended generation curve throughout the day, and
The solar array is heavily oversized relative to inverter (fronius) (in our case, approximately 90% oversizing).

As a result, DESS seems unable to correctly model or anticipate actual production behavior. It tends to underestimate early-day production from east-facing panels, which then leads to the battery reaching its maximum SOC too early in the day. This again causes production clipping and system inefficiency.

This mismatch could point to a limitation or flaw in the DESS forecasting or control logic, which may currently be optimized only for single-orientation or modestly sized PV systems.

Well. At least: VRM → Advanced → Solar Irradiance is pretty accurate.

I developer a PV Tracker flow that handles everything. Became pretty complex and now runs on a Raspberry Pi 5. This way I hope to be able to sell this as a product eventually (no need VenusOS Large).

The crux of the story is that I use a different provider to get data., and I am pretty pleased with it, thus far → forecast.solar

And yes. I do happen to have panels in six different orientations, with different tilt and azimuth settings. All to be setup before use,