Inverter matching line power

[raptor1956]

Some of the solar power systems for homes and the larger industrial scale systems convert the DC power from the solar panels to AC and then push that onto the grid when the output exceeds the local need.  Also, many small gen sets are inverter based and some permit the connection of two gens together to provide double the power.  In both cases the output of the inverter must be matched to the line power, or in the case of the double generator sets, to the other inverter, and I wonder how this is done.  Does a mcrocontroller monitor the grid power than match the frequency and adjust the voltage to allow power to be delivered. 

As a side point it would seem that a well designed system could, in fact, improve the quality of grid power by seeing where the waveform deviates from a perfect signwave and by outputting a carefully constructed waverform, improve the grid power. 


Brian

[jbb]

Hi Brian

Yes, typically a microcontroller is used for control.  Assuming a) a fair to good quality control scheme and b) a fairly normal power electronics design, the basic concept to drive power from an inverter into an existing grid looks like this:

• Line voltage is measured
• Software PLL locks onto line voltage fundamental (50 / 60 Hz) and strips off harmonics, noise etc.  It gives a continuous output of the line phase angle theta and (with some filtering) the average line voltage magnitude V_AC.
• A power reference P* comes in from somewhere (e.g. from solar MPPT control software)
• The average reference 'Direct axis' current ID* is calculated by ID* = P* / V_AC.  This is a DC quantity.
• The desired instantaneous current i* is generated using i* = ID* * cos(theta).  This is an AC quantity.
• A suitable current controller is used to adjust the PWM duty cycle such that the line current tracks the reference current.  Note: as this is an AC quantity, a standard PI controller isn't really suitable.  There are several options but they're quite technical.

This scheme doesn't try to control the line voltage - it just pushes some current into the grid.  The grid still controls the line voltage.


It is generally possible to play with the reference current to provide extra features.  For example, injecting reactive power can be added with the following modifications:

• A reactive power reference Q* comes in from somewhere
• The average reference 'Quadrature axis' current IQ* is calculated by IQ* = Q* / V_AC.  This is a DC quantity.
• The desired instantaneous current i* is generated using i* = ID* * cos(theta) + ID* * sin(theta).  This is an AC quantity and contains multiple components.

The downsides of this are: a) you have extra RMS current, and hence extra losses, and therefore don't make as much money selling power; and b) some inverter circuits aren't capable of it (or have special restrictions).


It is also possible to generate even more interesting current reference schemes, and indeed to provide 'tidy up' current to compensate for the behaviour of other devices.  A classic example is to tidy up harmonic currents from large thyristor drives (there are plenty still in service!).  There are again some downsides: a) extra RMS current again; and b) to get the best results you need high control speeds, which changes the power electronics requirements a bit and can decrease efficiency.


Note: I glossed over a lot with the 'suitable current controller', and indeed the whole bit of generating i* as an AC quantity. There are a lot of options, so I chose to present one which is conceptually simple.  If you want to get more into it, I can recommend:

• Looking at the design notes from Texas Instruments grid connected inverter reference designs.
• A book: Buso and Mattavelli, Digital Control in Power Electronics, Morgan and Claypool, 2006.  (Note: more recent editions may be available.)

[David Hess]

Another way to do it without a microcontroller or phase locking is to simply generate a reverse current proportional tho the line voltage.

Imagine a Howland current pump driving the line with a current proportional to the instantaneous voltage.  To make the output adjustable however, the instantaneous line voltage measurement needs to be multiplied by a variable factor so an analog multiplier is required to drive the Howland current pump.  The other input to the multiplier sets the output power which is very similar to how analog active power factor correction stages work.

Designing a high power Howland current pump is an exercise left to the reader but will likely be very similar to a class-D (switching) audio output stage; essentially you want a UPS or inverter output stage that is high impedance instead of low impedance and produces a controlled current instead of a voltage.