I'm new here - 1st post.
I did a bit of a search but didn't find much information about what other people have done to optimise internal power use of generated power.
Where I live this is important as exported power is paid out at 1/3 the rate of imported power so it's much more effective to use it than sell it back.
I already have a system so I though I'd share the configuration, reasoning behind the design and what I've done to maximise internal use of power.
I'm fortunate to be able to design and build most of the equipment I need for monitoring and while capable of designing and building my own inverters for grid-tie and AC/DC + DC/AC conversion for battery storage elected to go with off the self units since they were reasonably priced and meet all the standards requirements to get signed off.
I had 3 years of power use data prior to adding solar. I built my own power analyser to monitor the incoming mains and electric hot water heater. It's accurate to within +/- 0.1% and records the kJ used for every minute of the day. This gives me exact daily power use as well as when it was used. The minute data is stored into a PostgreSQL DB.
I got 4 years daily kWh generation data from a friend just down the road.
This allowed me to create a spreadsheet to dynamically model different sized systems and calculate cost vs payback.
The best payback for our power use and usage habits was:
24 x 260W Panel (6.24kW)
1 x SB5000TL Inverter
This system produces 45kWh on a good day.
I was getting about 15% more power than expected. I discussed with the guy down the road I modelled my system off and he said - oh yeah, I forgot to tell you, I get some partial shading at certain times of the day.. I guess this helps to account for some of the difference since I'm due North facing with no shading (Southern Hemisphere).
I then decoded to look at battery storage and it very quickly became apparent it was not cost effective. However, I went ahead anyway because what I had in mind meant an interesting engineering challenge.
8 x Trojan T-105
1 x Meanwell 2.4kW 48VDC Adjustable PSU
1 x Meanwell 0.6kW 12DC PSU (in series)
1 x Meanwell TS1500 Inverter
I ended up designing and building a control system based on the AtMega2560 microcontroller. Because this is a failry basic 8-bit uP with relatively low resolution ADC's and peripherals I added 16 bit ADC's (4) and 16 bit DAC's + a 16 Bit CPLD based phase control firing generator (EPM3064) to get reasonably high resolution data and control.
I use 2 of the ADC's to monitor the mains voltage and current and use the integral of the instanteaneous voltage x current over 2 seconds to get an accuate power reading even with relative high harmonic content. This value is used to create setpoints the the phase control and battery current.
I'm using phase control with a 0.1' accuracy (about 2W) to control the power to the hot water element. There's a filter between the SCR and mains to reduce distortion.
I can control the battery charge current to 10mA steps between +50A to -37.5A.
The PSU's charge the batteries.
The DC/AC inverter is always on with a server, firewall, network switches, fridge and a few other small load on a dedicated circuit in the house. It mean I could use a much cheaper non-grid-tie inverter.
A DCCT on the battery feed monitors the battery current and the control loop adjusts the PSU voltage to keep the battery charge discharge at or less than the setpoint current.
The control system gives priority to battery charging with any excess going into the hot water.
During the day our import/export power sits at zero (within +/- 0.010kW of zero) as long as:
1. There is enough solar energy to cover internal demand (otherwise we will be importing)
2. The batteries are charging (otherwise we could be exporting)
3. The hot water is less than max temp (otherwise we will be exporting)
When solar energy is available the batteries are charged using the 3-stage method; Constant current up the voltage corrected boost level (59.28V) then constant voltage until the current drops to 3% AH rating then float at 54.00V.
The setpoint current is reduced if solar energy is defficient in order to keep the import/export power on the mains as close to zero as possible.
I'm using simple amp-second state of charge prediction. Every 500ms I take the battery current, correct the discharge/charge current for Peukert's law and add/subtract from a running total. If 50% DOD is reached the discharge is stopped. If 95% state of charge is not acheived then discharge is inhibitted until it is reached. I get about 98% round trip accuracy with the error being under predicting the remaining amp-seconds. Once float is reached for 1 hour the amp-seconds are reset to maximum. When reset they are always within 2% through the modelling anyway.
The power is first maximised into battery charging and remaining power then into the hot water.
During the evening the battery setpoint current is -10 amps. This guarantees 12.5 hours minimum time on batteries before 50% DOD. If more than -10A is demanded the control loop injects current from the charging inverter to hold the battery current at -10A.
Using this method keeps our import/export power within +/- 0.010kW of zero for most of the day.
All power use, solar generation, battery charge state, charge current, battery temperatures and temperatures from 12 sensors on our hot water cylinder are stored every minute into a database. A Delphi based Windows app allows trending, reporting and real-time views of the history and status.
If anyone is interested in what I've done I'd be happy to send or publish the circuits, PCB's, C++ code and Delphi code (for Windows monitoring).
Is there anyone else interested in trying to optimise thier internal power use?
I did a bit of a search but didn't find much information about what other people have done to optimise internal power use of generated power.
Where I live this is important as exported power is paid out at 1/3 the rate of imported power so it's much more effective to use it than sell it back.
I already have a system so I though I'd share the configuration, reasoning behind the design and what I've done to maximise internal use of power.
I'm fortunate to be able to design and build most of the equipment I need for monitoring and while capable of designing and building my own inverters for grid-tie and AC/DC + DC/AC conversion for battery storage elected to go with off the self units since they were reasonably priced and meet all the standards requirements to get signed off.
I had 3 years of power use data prior to adding solar. I built my own power analyser to monitor the incoming mains and electric hot water heater. It's accurate to within +/- 0.1% and records the kJ used for every minute of the day. This gives me exact daily power use as well as when it was used. The minute data is stored into a PostgreSQL DB.
I got 4 years daily kWh generation data from a friend just down the road.
This allowed me to create a spreadsheet to dynamically model different sized systems and calculate cost vs payback.
The best payback for our power use and usage habits was:
24 x 260W Panel (6.24kW)
1 x SB5000TL Inverter
This system produces 45kWh on a good day.
I was getting about 15% more power than expected. I discussed with the guy down the road I modelled my system off and he said - oh yeah, I forgot to tell you, I get some partial shading at certain times of the day.. I guess this helps to account for some of the difference since I'm due North facing with no shading (Southern Hemisphere).
I then decoded to look at battery storage and it very quickly became apparent it was not cost effective. However, I went ahead anyway because what I had in mind meant an interesting engineering challenge.
8 x Trojan T-105
1 x Meanwell 2.4kW 48VDC Adjustable PSU
1 x Meanwell 0.6kW 12DC PSU (in series)
1 x Meanwell TS1500 Inverter
I ended up designing and building a control system based on the AtMega2560 microcontroller. Because this is a failry basic 8-bit uP with relatively low resolution ADC's and peripherals I added 16 bit ADC's (4) and 16 bit DAC's + a 16 Bit CPLD based phase control firing generator (EPM3064) to get reasonably high resolution data and control.
I use 2 of the ADC's to monitor the mains voltage and current and use the integral of the instanteaneous voltage x current over 2 seconds to get an accuate power reading even with relative high harmonic content. This value is used to create setpoints the the phase control and battery current.
I'm using phase control with a 0.1' accuracy (about 2W) to control the power to the hot water element. There's a filter between the SCR and mains to reduce distortion.
I can control the battery charge current to 10mA steps between +50A to -37.5A.
The PSU's charge the batteries.
The DC/AC inverter is always on with a server, firewall, network switches, fridge and a few other small load on a dedicated circuit in the house. It mean I could use a much cheaper non-grid-tie inverter.
A DCCT on the battery feed monitors the battery current and the control loop adjusts the PSU voltage to keep the battery charge discharge at or less than the setpoint current.
The control system gives priority to battery charging with any excess going into the hot water.
During the day our import/export power sits at zero (within +/- 0.010kW of zero) as long as:
1. There is enough solar energy to cover internal demand (otherwise we will be importing)
2. The batteries are charging (otherwise we could be exporting)
3. The hot water is less than max temp (otherwise we will be exporting)
When solar energy is available the batteries are charged using the 3-stage method; Constant current up the voltage corrected boost level (59.28V) then constant voltage until the current drops to 3% AH rating then float at 54.00V.
The setpoint current is reduced if solar energy is defficient in order to keep the import/export power on the mains as close to zero as possible.
I'm using simple amp-second state of charge prediction. Every 500ms I take the battery current, correct the discharge/charge current for Peukert's law and add/subtract from a running total. If 50% DOD is reached the discharge is stopped. If 95% state of charge is not acheived then discharge is inhibitted until it is reached. I get about 98% round trip accuracy with the error being under predicting the remaining amp-seconds. Once float is reached for 1 hour the amp-seconds are reset to maximum. When reset they are always within 2% through the modelling anyway.
The power is first maximised into battery charging and remaining power then into the hot water.
During the evening the battery setpoint current is -10 amps. This guarantees 12.5 hours minimum time on batteries before 50% DOD. If more than -10A is demanded the control loop injects current from the charging inverter to hold the battery current at -10A.
Using this method keeps our import/export power within +/- 0.010kW of zero for most of the day.
All power use, solar generation, battery charge state, charge current, battery temperatures and temperatures from 12 sensors on our hot water cylinder are stored every minute into a database. A Delphi based Windows app allows trending, reporting and real-time views of the history and status.
If anyone is interested in what I've done I'd be happy to send or publish the circuits, PCB's, C++ code and Delphi code (for Windows monitoring).
Is there anyone else interested in trying to optimise thier internal power use?
Comment