I have an VRLA battery bank of 48V/533aH. They are charged by 3 Victron MPPT charge controllers. Measurement is via a SmartShunt connected to both the Sunny Islands and the Cerbo. The inverters are 2 Sunny Island SI08 and SI06 set as master/slave. Reporting is via a Cerbo GX (II). The system is Offgrid.
In a recent event of rain and heavy clouds the batteries were not able to charge completely for a 4 day period. I have attached a screenshot from the VRM portal.
On the fourth day I manually synchronised the SOC to 100% when the Cerbo was indicating a SOC of 92% (green line on graph). The MPPTs were all on float.
When I download the VRM data which is recorded every minute and analyse the results. The system correctly identifies the aH consumed and the aH charged, so that at the end of the 4 day period the data indicated a deficit of 48 aH. By way of example 48aH @ 49V is a 2352 Watt deficiet or 9.2% undercharged
When I ran the data through a spreadsheet and use the Shunt data, (Volts multiplied by the Amps to give me watts), the cumulative data indicates that 22,717 watts were consumed and 24358 watts were charged to the system. This basic analysis is in line with the MPPT’s being in float. The battery is overcharged by 7% which will the the efficiency losses of a standard VRLA system. My historical losses are more in the 10-11% range, but this is simply a splitting of hairs when I am only using 4 days of data.
The mathematical anomaly seems to relate to drawing power at voltages predominately below 51V and yet charging voltages are generally well in excess of 55V. If I discharge at a constant voltage and recharge at the same constant voltage then the aH excluding losses would be equal if the battery goes from full charge to full charge. However when I discharge and charge at different voltages the aH consumed/charged start to vary
Is there a method of reflecting a more accurate SOC, in light of the above scenario? Why does the Cerbo GX use the aH consumed/charged as the primary calculation of SOC and not the Watts consumed/charged which to my simple logic would appear to be more accurate. (am leaving aside the issues of Peukert, efficiency looses and tail current for the purposes of simplicity of this argument).