[RE-wrenches] utility line voltage issues

[Bill Brooks]

Peter,

The excursion time depends on how high the excursion is. There is a longer
grace period for being just above or below the limits to allow for inverters
to ride through normal transients. However, beyond that are some very short
timeframes for excessive transients and large excursions. There is a table
in 1547 that covers this. This is the table that is used for UL1741 testing.
The "instant trip" rates are typically 2 cycles to correspond to fast
operating relays. An inverter can trip as fast as you want it to. The
question is how much information you accumulate before you realize you are
in a situation where a trip is required.

Every manufacturer is required to pass the voltage tests, but there is no
requirement to ride through the full grace period and then trip. This causes
some inverters to overreact to voltage excursions and transients that the
standard intended to allow inverters to ride through. In that case, the
inverter manufacturer would be to blame for being overly sensitive. Lots of
blame to go around if you want to play the blame game.

A grid-connected inverter is regulating output power by measuring line
voltage and determining how much current to inject for a given voltage
value. Looking at the trend of the last few voltage cycles, it predicts what
the next voltage cycle voltage will look like (magnitude) and injects the
appropriate amount of current that is available with power from the PV
array. For example, a PV array has enough power to produce an ac output of
4,000 Watts (meaning it has 4,200Watts dc on an efficient inverter). The ac
voltage at the inverter terminals was 245Vac for the past few cycles, so the
inverter produces a perfect current waveform of 16.33A ac (I = P/V =
4,000W/245V = 16.33A). Now let's say that on the next cycle all of a sudden
the voltage drops to 50%. The inverter would normally try to double the
current, so to prevent damage to the transistors, the inverter limits the
current and will either stop producing current shut down immediately. You
mention grid impedance, but the inverter may not know the actual grid
impedance until it has produced its first few cycles. The voltage will rise
and the inverter must compensate--up to the upper voltage limit of the
inverter (+8% above nominal).

Bill.



-----Original Message-----
From: re-wrenches-bounces at lists.re-wrenches.org
[mailto:re-wrenches-bounces at lists.re-wrenches.org] On Behalf Of Peter
Parrish
Sent: Monday, August 17, 2009 11:00 AM
To: 'RE-wrenches'
Subject: Re: [RE-wrenches] utility line voltage issues

One thing I have read during this thread, is the voltage rise at the
inverter output required to "sell to the grid". I have heard all sorts of
numbers from "less than a volt RMS" to "2-3 volts". As an engineer, I would
expect the number to depend on the AC current being exported and the
impedance of the grid seen from the distribution panel.

Let's say we want to export 2,400 W-ac at 240 V-ac, or 10 A-ac, and the
dynamic impedance of the grid is 0.1 ohm. The voltage rise needed to push
that 10 A onto the grid is 1 V-ac (i.e. 241 V-ac).

I am going to do a test on our system today, monitoring the L1-L2 voltage at
the inverter breaker and cycle the PV system on and off. Should be
interesting! 

One other point: since IEEE929 and IEEE1547 require the inverter to shut
down within 16 ms (one full period of 60 Hz) -- correct? -- what is the
typical averaging time (Tavg) for a typical RMS voltmeter (the "M=mean" in
RMS)? If the voltage excursion occurs between 0.16mS and a fraction of Tavg,
then the RMS will miss the event. Seems to me, we have to have a peak
reading voltmeter or something equivalent.

- Peter