Inverter design - reactive loads

jonr
jonr Solar Expert Posts: 1,386 ✭✭✭✭
I'm curious, when a non-grid-tie sine-wave inverter is connected to a reactive load, the load doesn't consume all of the energy, it stores it and tries to return it to the source.   So when returned to the source (the inverter), where exactly does it go?   My understanding is that inverters that can actually return energy all the way back to the battery are rare (diode rectifiers are one-way).   Looking at the below circuit diagram and picking some worse case reactive loads, my guess is:

a) a 100% capacitive load would cause a squaring of the trailing edge of the sine wave (ie, voltage couldn't decline) and, just after the zero crossing, very large current flows from the load to C3.  Certainly causing heating of the mosfets and possibly causing a reversal of the polarity on C3.  The energy is wasted as mosfet heat with no energy being recovered.

b) a 100% inductive load would generate back voltages above the normal AC waveform peak and apply it to C3.  The energy is stored there and could be used on the next cycle (or, if there is nothing to consume it, it would keep building up).

Both cases appear dangerous to C3.

http://solar.smps.us/grid-tie-inverter-schematic.html


I am available for custom hardware/firmware development

Comments

  • BB.
    BB. Super Moderators, Administrators Posts: 33,433 admin
    I am not an inverter designer--But (at least for sine wave inverters), the energy flow can flow all the way back to the battery bank.

    And if you drive an off grid sine wave inverter would enough voltage on its output, you will have full power flow reversal and actually charge the battery bank. This has been a "feature" (mostly undocumented, but now some vendors do document) for AC Wind Turbines and GT Solar Inverter systems to take them off grid.

    This "feature" (backflow of energy) has now been implemented in some AC OG inverters with a method to limit charging current to the battery bank (a standard OG inverter will charge the battery bank until the inverter hits high bank voltage alarm or something fails)--The "retrofit way" was to add a battery voltage sensor to send a signal to a relay to cut the power from the GT inverter to stop charging (5 minute timeout on the GT inverter). Another is to have the sine wave inverter start to modify its output frequency to +/- 1 Hz or or so--This will cause the GT inverter to "unsync" with the OG inverter and shut down until the line frequency is within specifications. As I understand, the line frequency is cycled between "high and low" limits--This will help AC timers keep accurate time.

    The reason this works? AC inverters do not use diodes to control power flow in the circuit. They use something similar to H Bridge switching. The "transformer" is the middle of the H (horizontal bar). The 4 vertical bars are each a transistor/FET switch that is controlled by electronics to cycle on/off on command.

    The battery + at the "top" and battery - at the "bottom. Top Left and Bottom Right on, current flows one direction through the transformer (H Bar). Then Top RIght/Bottom Left, and you have reverse current flow. The other winding of the transformer is your high voltage AC output.

    So, what you have now is a "synchronous" rectifier system. If you supply power backwards through the H Bar "Transformer", then the current will recharge the battery bank as the FET switches are not diodes by bi-directional switches that allow current to flow either way. And the electronics control the timing "in phase" with the AC voltage (or control the AC voltage/wave form in standard AC inverter function). It does "look" like a diode bridge, but it is not. FETs also are more efficient than diodes (no diode voltage drop), so synchronous rectifiers do not have the diode losses common to a "simple" bridge rectifier.

    Note that sine wave inverter outputs are way more complicated than H Bridge circuits--The inverter needs to vary the current/voltage profile to generate a "95% pure" sine wave.

    Again--this is not my field--So my explanations are pretty simple and may boarder on "wrong", in some respects, when you get into the details of a real inverter and AC power math.

    -Bill
    Near San Francisco California: 3.5kWatt Grid Tied Solar power system+small backup genset
  • dennis461
    dennis461 Registered Users Posts: 109 ✭✭✭
    jonr said:
    I'm curious, when a non-grid-tie sine-wave inverter is connected to a reactive load, the load doesn't consume all of the energy, it stores it and tries to return it to the source.    
    .....


    Unfortunately, you started with an incorrect statement.  The grid, load, whatever, does not try to return anything.  Reactive power (sometimes called apparent power) is wasted by the load, transmission lines.  There are plenty of good papers describing the situation. It is lost. 
    Camden County, NJ, USA
    19 SW285 panels
    SE5000 inverter
    grid tied
  • jonr
    jonr Solar Expert Posts: 1,386 ✭✭✭✭
    edited March 2017 #4
    So the summary is that the D1-D5 diodes shown in the example design aren't there in most or some actual OG designs (instead using switched FETs or perhaps a design that doesn't use a high voltage DC link/bus).  I suppose such regeneration is easy to check - put an inductive load on a OG inverter while measuring current flow from the battery.   Use a scope to look for any brief reversals.

    I expect that such inverters would show cleaner waveforms with reactive loads.  And be more efficient.

    In the example circuit, the output H-bridge does fast PWM - allowing it to produce a sine wave.

    I am available for custom hardware/firmware development

  • BB.
    BB. Super Moderators, Administrators Posts: 33,433 admin
    OK--I missed your link.

    That is a circuit for a GT (grid tied) inverter. The first Q1 is part of a "boost" converter. It raises/regulates the variable DC voltage+current from the solar array (solar panels are definitely "one way" power flow devices--The will not accept high back voltage/current without damage). And D1 makes the "boost converter" a one way power flow device too.

    The first stage H Bridge is nothing more than taking your "low voltage" solar array input voltage and jacking it up to ~373 VAC (rough "intermediate DC voltage" to drive 240 VAC nominal inverter) for C3. Q2-Q5 are bi-directional (regarding power flow) or a synchronous AC inverter.  But since boost converter is one way--reverse power flow would over-volt C2 (boost converter output stage).

    Q6 through Q9 could support reverse power flow too... However, a typical GT inverter operates in an entirely different "regulation mode" vs an Off Grid inverter.

    An OG Inverter is a "voltage regulated AC source". The control circuitry is designed to hold 120 VAC (or 240 VAC) regardless of load. Normally your "loads" only consume power, they do not generate power, so the OG inverter has no circuitry to prevent damage from reverse current flow (basically over voltage/over charging the DC battery bank). An OG Inverter has very close to Zero Ohms input resistance (near zero Ohms is the resistance of a "good" battery).

    A typical GT inverter runs as a "current regulated source". A current regulated to output (for example) 10 amps regardless of the output voltage and has near infinite output resistance (i.e., if you tried to feed current backwards through a current source, no current will flow until you "blow something up").

    A GT inverter simply "tracks the AC sine wave voltage" and outputs current proportional to the sine wave voltage. At zero volt crossing it outputs 0 amps and at peak AC voltage it outputs 10 amps (240 VAC * sqrt(2) = 339 Volts peak of sine wave). [---Really the math is a bit more complex--The way "average current" is measured is RMS--Root Mean Square, so the 240 VAC current would be 10 amps and the 339 volt peak current would be ~14 amps peak].

    The fact that the controller's electronics (negative feedback, control theory, etc.) can change the output of a switching power supply so completely (from voltage source to current source) has always been amazing to me.

    A car analogy... You drive (typically) to the speed limit. You control the gas for the engine to keep 65 MPH on the flats, on the hills, head/tail winds, etc.... That is "MPH Regulation". If you have a gasoline engine car, you need to use the brakes if going down a steep hill to keep the speed limit. If you have an electric/hybrid car, the extra energy can recharge the battery bank (via H Bridge electronics).

    On the other hand, going from 0 to 65 MPH--Say you have a sick kid or fragile load and you want even acceleration. For example, you control the accelerator pedal to increase speed by 1 MPH per second regardless of speed--Until you get up to 65 MPH, then you converter to MPH Regulation--Basically how an MPPT charge controller works by the way. Going up a hill, more gas to maintain 1 MPH/Sec increase. Going down hill, less gas/more brake to not let the speed increase too quickly--"Current Regulation" mode).

    Anyway--If you measure the DC current to an AC OG inverter--You will see a "sine squared" current wave form (120 Hz from near zero to peak current--sine wave like profile). There is no energy storage in an OG AC inverter--All of the fundamental frequency current/energy most come from and go back to the battery bank.

    I have not done the experiment, but if you put a pure inductor on the output of the Sine Wave inverter (assuming the inverter works correctly and does not go unstable or shut down due to pure inductive load), I would expect to see bidirectional current flow with the battery bank. During 1/2 the cycle, there will be current flow from the battery bank, and the other 1/2 of the cycle, current going back into the battery bank (excluding losses). Very much like you pushing and pulling on a weight on tracks--You push to accelerate the weight and pull to slow the weight down.

    Similar to if you put a capacitor on the output of the AC inverter--Now you are charging/discharging a capacitor. (180 degree opposite phase to inductor current wave form). Just like pushing on a spring--compressing/decompressing.

    There are synchronous GT inverters too. This are pure voltage regulators in GT mode and must synchronize with the AC voltage wave form. Also, paralleling AC Gensets (large installations like hospitals) use synchronizing voltage sources to share AC loads and (during genset weekly/monthly testing) can be configured to feed energy back to the utility mains (use the AC mains as the test load instead of wasting energy in a resistor bank). Connect a large genset to the AC mains at the wrong frequency/out of phase--You can have big problems on your hands.

    There are many things that are bi-directional but we do not even think about it... For example, take an AC induction motor (single phase, 3 phase, does not matter). And attach it to a gasoline motor. Turn the electric motor one (AC mains), the induction motor will turn the gasoline motor ("normal"). Now, on the gasoline motor, supply fuel and spark, and start applying throttle. At the point the gasoline motor starts supplying torque to the induction motor, you have now turned the AC motor into an AC generator that is "sync'ed" to the grid voltage. You can now feed energy backwards to the utility and turn your utility meter backwards (if you have an older utility meter).

    People have used this function for small hydro. Get the turbine+AC motor up to speed, then turn on the water to drive the AC motor, and feed power back into the AC mains (or off grid sine wave AC inverter).

    Here is a very nice set of videos that really show how motor/generators and AC power (three phase in this case) works. An electrical engineering 101 lab example:

    https://www.youtube.com/watch?v=RGPCIypib5Q
    https://www.youtube.com/watch?v=sFohkp2aaU4
    https://www.youtube.com/watch?v=GRk_qJxaxh8
    https://www.youtube.com/watch?v=RT1ySBc-Bls

    Real life example of test exercising 500 kVA genset with grid as load (all computer controlled):

    https://www.youtube.com/watch?v=SS_aHWdLH54

    Note the variation of terms... Watts and kWatts vs VA and kVA. Power or energy in measured in Watts and Watt*Hours (kW, kWH). Pure resistive loads with Power Factor of 1.0

    VA is what you are asking about--Inductive and capacitive loads. Motors are usually inductive loads

    Utilities charge residential customers for Watts and Watt*Hours (kWH).

    Utilities charge large commercial customers in kWH, but they also take into account kVA--If a customer has a bunch of induction motors (oil refinery), the motors could be running a power factor of 0.67 (pretty bad). The utility will charge the customer a penalty of 1/0.67 * their kWH costs (1/0.67 = 1.49x their "energy usage"). For the utility, they have to size the wiring, transformers, and even the utility generators to send out 1.49x more VA (kVA) vs what the customer is using in Watt/kWH. So--Large customers will design their installations to keep PF near 1.0 (actually 0.95 for other reasons)--So they do not get hit with a "power factor" penalty. For induction motors, this is done by adding parallel capacitors (capacitor+inductor of the the right size in parallel give PF~1.0).

    I will stop here--There is a lot of information to take in.

    -Bill





    Near San Francisco California: 3.5kWatt Grid Tied Solar power system+small backup genset
  • jonr
    jonr Solar Expert Posts: 1,386 ✭✭✭✭
    edited March 2017 #6
    Thanks Bill.  I see some perhaps more typical OG designs here: http://solar.smps.us/sinewave-inverters.html
    But note the comment "Sometimes a boost converter is added before the isolating stage".  Also here: http://www.dieselduck.info/machine/04 auxiliary/2000 Inverter technology.pdf

    Apparently DC-link over-voltage is a common problem in diode rectifier based VFD designs (when the motor turns into a generator).  Their solution is a braking resistor.

    >  The grid, load, whatever, does not try to return anything...It is lost. 

    As Bill explains, reactive loads do return energy to the source.  Very little is lost (only due to some additional wire heating).  Let me know if you have a link that you think says otherwise.

    I am available for custom hardware/firmware development

  • jonr
    jonr Solar Expert Posts: 1,386 ✭✭✭✭
    Looking at some typical boost converter designs,  it looks like they would require modifications to be able to efficiently return energy to the battery (ie, act as a buck converter in the reverse direction when the output voltage is too high).

    I am available for custom hardware/firmware development