Asymmetrical Waveform Generation for Fault Testing of Grid Connected

Solar Inverters

Abstract
This paper discusses the calculations for generating various asymmetrical grid voltage conditions in a test
laboratory environment to evaluate distributed generation units, particularly three-phase solar inverters
that interconnect with an external transformer. The laboratory utilizes a four-wire, wye-grounded 4-
quadrant AC source, or grid simulator, with an external voltage reference input. In this case, the voltage
profile is generated through a sequencer to adjust amplitudes and phases for each phase-ground voltage.
Derived calculations are used to determine representative phasors for various asymmetrical faulted
voltages for different distributed generation unit connections. Implications on instantaneous waveform
generation are then considered for the faulted phase(s) as well as the unfaulted phase(s).

i. Introduction
Solar photovoltaic (PV) systems interface with three-phase power systems through inverters and
transformers. Three-phase solar inverters typically have a low voltage output, in either three-wire or four-
wire configurations, and are coupled to a transformer to interconnect with medium voltage collection
systems or distribution networks. Transformer winding specifications are often determined by the PV
array grounding method with ungrounded arrays calling for grounded transformer secondaries and
grounded arrays specifying floating secondaries. Beyond this specification, and perhaps a minimum
impedance, the inverter manufacturer transformer requirements are relatively broad. However, the
number of wires (three or four), along with the transformer winding configuration ultimately determine
how distribution and transmission system faults propagate through to the internal inverter voltage
measurements. Increased penetration of distributed generation onto the power system has led to more
requirements on distributed generation units to support bulk power system stability such as riding
through grid faults [Rule 21, P1547 Draft].

Recent wildfires in California revealed the potential for tripping of large amounts of inverter-connected
generation as a result of grid faults.[X]

Although a three-phase symmetrical fault may appear to be the worst case and most important to
mitigate against, the occurrence of line-to-line and line-to-ground faults are much more common. [X]

In the test laboratory environment, the inverter is connected to a simulated utility, or grid simulator, for
grid code evaluation which includes asymmetrical voltage condition tests [X].

There are several aspects of asymmetrical voltage, or fault, generation to consider for inverter evaluation
in a testing environment. The first is the type of simulated utility, or grid simulator, including the topology
which affects whether it should be considered a three or four wire source and may also play a role in
whether the grid simulator is natively low or medium voltage. If the grid simulator is setup for medium
voltage output natively then a representative medium voltage transformer can be used to interconnect
the inverter using the manufacturer recommendations. For a low voltage grid simulator, the inverter may
be connected directly, provided that its grounding configuration is compatible with the rest of the
experimental setup, or through a low voltage to low voltage transformer with representative winding
configuration. The second aspect is the determining which voltage bus is the point of interest (i.e. where
the representative asymmetrical voltage should be generated), either the inverter terminals or the grid
simulator side of an interfacing transformer. This is important as it determines which types of asymmetries
should be included in the design of experiments. Once these aspects are determined, the grid simulator
needs to be programmed to generate the different types of asymmetrical faults so that the inverter
response can be evaluated.

This paper will focus on experiment setups with an interfacing transformer and will cover cases where the
point of interest is either at the inverter terminals or at the grid-side transformer terminals. Depending
on the transformer connection, the asymmetrical condition generated by the grid simulator may not be
the same as the one seen in the terminal of the inverter. So, in order to create a specific unbalance
condition at the inverter terminal, a set of equations has to be solved for each transformer connection
scheme to calculate the required grid side voltage. These calculated voltages should then be generated
by a grid simulator to result in a correct unbalanced condition at the terminal of the device under test.

In this paper, the process to obtain the grid-side voltage based on the inverter terminal voltage will be
detailed with some numerical examples. A circuit simulation will be presented to validate the calculations.
Finally, the experimental results of applying asymmetrical conditions on a megawatt scale inverter with
voltage ride-through functions will be given to confirm the process.

ii. Asymmetrical Fault Generation

The essence of the problem is that the inverter is connected to power amplifier unit (PAU) via the line to
line secondary windings of a Δ-Y transformer as shown in figure X1. To further complicate matters, the
line-to-neutral voltages of the PAU are what are controllable. The question essentially becomes what
line-to-neutral voltages are required on the primary of the Δ-Y transformer in order to recreate the
desired line-to-line fault voltages on the secondary?

Power Amplifier Unit Solar Array Inverter

Va
TX1
1u
ACMAG = {1*2400}
ACPHASE = -60

15:1

V_AB
TX2 Vb
1u
ACMAG = {1*2400}
ACPHASE = -180
0 V_CA

15:1
V_BC

Vc
TX3
1u
ACMAG = {1*2400}
ACPHASE = 60

15:1

1MEG 0

Starting with the desired inverter line-to-line voltages and working back towards the PAUs, the first step
is to find the necessary line-to neutral voltages on the Y-connected secondary of the transformer. This
can be done utilizing a geometric solution similar to that utilized in traditional voltage sag analysis
{Bollen}. It is also possible using symmetric components, which will be discussed in a subsequent paper.
In order to find a unique solution, it is assumed that a three-wire system is used, thus ensuring there is
no neutral current. From a geometric standpoint, this means that the vector sum of both the line-to-line
voltages and the line-to-neutral voltages are zero.

Using this assumption, equations A through F establish the relationships between line-to-line and line-
to-neutral voltages for both balanced and unbalanced conditions. These relationships are shown
geometrically in figures X2a and X2b for a balanced system. In all of the figures presented, the angle of
VAB is used as a reference and the maximum line-to-line voltage, whether VAB or VBC, is kept at 1p.u.

1
⃗⃗⃗⃗⃗⃗ ⃗⃗⃗⃗⃗⃗
𝑉𝐴𝐵 = (𝑉 ⃗⃗⃗⃗⃗⃗⃗
𝐴𝑁 − 𝑉𝐵𝑁 ) [A] ⃗⃗⃗⃗⃗⃗
𝑉 ⃗⃗⃗⃗⃗⃗ ⃗⃗⃗⃗⃗⃗
𝐴𝑁 = 3 (𝑉𝐴𝐵 − 𝑉𝐶𝐴 ) [D]
1
⃗⃗⃗⃗⃗⃗ ⃗⃗⃗⃗⃗⃗⃗
𝑉𝐵𝐶 = (𝑉 ⃗⃗⃗⃗⃗⃗
𝐵𝑁 − 𝑉𝐶𝑁 ) [B] ⃗⃗⃗⃗⃗⃗⃗
𝑉𝐵𝑁 = 3 (𝑉⃗⃗⃗⃗⃗⃗ ⃗⃗⃗⃗⃗⃗
𝐵𝐶 − 𝑉𝐴𝐵 ) [E]
1
⃗⃗⃗⃗⃗⃗ ⃗⃗⃗⃗⃗⃗
𝑉𝐶𝐴 = (𝑉 ⃗⃗⃗⃗⃗⃗
𝐶𝑁 − 𝑉𝐴𝑁 ) [C] ⃗⃗⃗⃗⃗⃗
𝑉𝐶𝑁 = (𝑉 ⃗⃗⃗⃗⃗⃗ ⃗⃗⃗⃗⃗⃗
𝐶𝐴 − 𝑉𝐵𝐶 ) [F]
3

Under a line-to-ground fault, one of the line-to-line voltages will decrease by some percentage. For
simplicity, we will assume this is VAB. This will mean that VBC and VCA must decrease their phase angles to
compensate for the reduced amplitude of VAB. Figures X3a and X3b show the geometry as VAB is
decreased to 70% of its original value while the magnitudes of VBC and VCA remain at 1p.u.
Under a line-to-line fault, two of the line-to-line voltages will decrease by some percentage, while the
third will stay at its original level. Assuming VAB stays at its original level, VBC and VCA must increase their
phase angles to compensate. Figures X4a and x4b show the geometry for voltages VBC and VCA with a
reduction to 70% in magnitude. The depth of this sag cannot be less than 50%.

Because of the interdependency of line-to-line voltages, if VAB is kept as a reference, the real portions of
VBC and VCA must equal half of the magnitudes of VAB but in the opposite direction and the imaginary
portions of VBC and VCA must be equal to each other but opposite in magnitude. These relationships are
shown in Equations G through I. In these equations, Vo is the original voltage magnitude and x is the
depth of the sag, expressed as a fraction of the original voltage. Combining these equations with the
relationships established in equations D-F, the line-to-neutral voltages can be expressed as shown in
equations J-L.

𝑥 𝑗 𝑥 2
⃗⃗⃗⃗⃗⃗
𝑉𝐴𝐵 = 𝑥𝑉𝑜 + 𝑗0 [G] ⃗⃗⃗⃗⃗⃗
𝑉 𝐴𝑁 = 𝑉𝑜 (2 − 3 √1 − (2 ) ) [J]

−𝑥 𝑥 2 −𝑥 𝑗 𝑥 2
⃗⃗⃗⃗⃗⃗
𝑉𝐵𝐶 = 𝑉𝑜 ( 2 − 𝑗√1 − (2) ) [H] ⃗⃗⃗⃗⃗⃗⃗
𝑉𝐵𝑁 = 𝑉𝑜 ( 2 − 3 √1 − (2) ) [K]

−𝑥 𝑥 2 2𝑗 𝑥 2
⃗⃗⃗⃗⃗⃗
𝑉𝐶𝐴 = 𝑉𝑜 ( 2 + 𝑗√1 − (2) ) [I] ⃗⃗⃗⃗⃗⃗
𝑉𝐶𝑁 = 𝑉𝑜 ( 3 √1 − (2) ) [L]
For a fault where VBC and VCA are reduced in amplitude, the equations are nearly identical as shown in
equations M-R.

1 𝑗 1 2
⃗⃗⃗⃗⃗⃗
𝑉𝐴𝐵 = 𝑉𝑜 + 𝑗0 [M] ⃗⃗⃗⃗⃗⃗
𝑉 𝐴𝑁 = 𝑉𝑜 (2 − 3 √𝑥 − (2) ) [P]
2

−1 1 2 −1 𝑗 1 2
⃗⃗⃗⃗⃗⃗
𝑉𝐵𝐶 = 𝑉𝑜 ( 2 − 𝑗√𝑥 2 − (2) ) [N] ⃗⃗⃗⃗⃗⃗⃗
𝑉𝐵𝑁 = 𝑉𝑜 ( 2 − 3 √𝑥 2 − (2) ) [Q]

−1 1 2 2𝑗 1 2
⃗⃗⃗⃗⃗⃗
𝑉𝐶𝐴 = 𝑉𝑜 ( + 𝑗√𝑥 2 − ( ) ) [O] ⃗⃗⃗⃗⃗⃗
𝑉𝐶𝑁 = 𝑉𝑜 ( √𝑥 2 − ( ) ) [R]
2 2 3 2

Now that the relationship of the required line-to neutral voltages for the secondary of the Δ-Y
transformer have been established based on the fault type and depth, it is a simple matter of repeating
the process on the primary side of the transformer. For a Δ-Y transformer with line-to-line voltage ratios
given by 𝑎, the primary line-to-line voltages are in phase with the secondary line-to-neutral voltages but
are scaled by √3𝑎. These line-to-line voltages are given for single phase sags in equations S-U and for
two phase sags in equations V-X.

𝑥 𝑗 𝑥 2 1 𝑗 1 2
⃗⃗⃗⃗⃗⃗⃗⃗
𝑉𝐴𝐵′ = √3𝑎𝑉𝑜 ( − √1 − ( ) ) [S] ⃗⃗⃗⃗⃗⃗⃗⃗
𝑉𝐴𝐵′ = √3𝑎𝑉𝑜 ( − √𝑥 2 − ( ) ) [V]
2 3 2 2 3 2

−𝑥 𝑗 𝑥 2 −1 𝑗 1 2
⃗⃗⃗⃗⃗⃗⃗⃗
𝑉𝐵𝐶 ′ = √3𝑎𝑉𝑜 ( 2 − 3 √1 − (2) ) [T] ⃗⃗⃗⃗⃗⃗⃗⃗
𝑉𝐵𝐶 ′ = √3𝑎𝑉𝑜 ( 2 − 3 √𝑥 2 − (2) ) [W]

2𝑗 𝑥 2 2𝑗 1 2
⃗⃗⃗⃗⃗⃗⃗
𝑉 𝐶𝐴′ = √3𝑎𝑉𝑜 ( 3 √1 − (2 ) ) [U] ⃗⃗⃗⃗⃗⃗⃗
𝑉 𝐶𝐴′ = √3𝑎𝑉𝑜 ( 3 √𝑥 − (2) )
2 [X]

Finally, using equations D-F, the line to neutral voltages of the PAUs can be calculated as shown in
equations Y-DD. Equations Y-AA are for a single phase sag and BB-DD are for two phases.

𝑎𝑉 𝑥 𝑥 2 𝑎𝑉 1 1 2
⃗⃗⃗⃗⃗⃗⃗⃗
𝑉𝐴𝑁′ = 3𝑜 (2 − 𝑗√1 − (2) ) [Y] ⃗⃗⃗⃗⃗⃗⃗⃗
𝑉𝐴𝑁′ = 3𝑜 (2 − 𝑗√𝑥 2 − (2) ) [BB]
√ √
−𝑎𝑉𝑜𝑥 −𝑎𝑉𝑜
⃗⃗⃗⃗⃗⃗⃗⃗
𝑉𝐵𝑁′ = + 𝑗0 [Z] ⃗⃗⃗⃗⃗⃗⃗⃗
𝑉𝐵𝑁′ = + 𝑗0 [CC]
√3 √3
𝑎𝑉𝑜 𝑥 𝑥 2 𝑎𝑉𝑜 1 1 2
⃗⃗⃗⃗⃗⃗⃗⃗
𝑉 𝐶𝑁′ = 3 (2 + 𝑗√1 − (2 ) ) [AA] ⃗⃗⃗⃗⃗⃗⃗⃗
𝑉 𝐶𝑁′ = 3 (2 + 𝑗√𝑥 − (2) ) [DD]
2
√ √

iii. Simulation and Experimental Results

 Simulation results of Simscape Power Systems (Can it be PLECS or PSpice?) on both cases
(Y-Y and ∆-Y) discussed in section ii.
 Experimental setup and results for one the cases
The results were found through programming in Simulink as well, using Simscape Power Systems blocks.
Utilizing (Y-DD), phase voltages for the PAU units are created through AC voltage sources, specifying the
voltage and phase of the respective phase. The voltages are connected to a three phase transformer and
terminated by voltage measurement blocks. The simulation provides outputs of RMS voltages and phase
angles for both points of interest on the sides of the transformer as line-to-neutral and line-to-line values.

Figure 5. Output for a line-to-ground fault

Figure 6. Output for a line-to-line fault

iv. Conclusion
  Comparison of simulation and experimental case
 Future work?
o Symmetrical components
o Simulation consolidation to Simscape
References:
IEEE1547:

-“IEEE Standard for Interconnecting Distributed Resources with Electric Power

Systems,” in IEEE Std 1547-2003 , vol., no., pp.1-28, July 28 2003
doi: 10.1109/IEEESTD.2003.94285
UL1741:
EIC paper:
eGRID papers:

J. C. Fox and B. Gislason, "Introduction to the Clemson University 15 MW Hardware-

In-the-Loop Grid Simulator," 2014 Clemson University Power Systems Conference,
Clemson, SC, 2014, pp. 1-5.
doi: 10.1109/PSC.2014.6808118