SPARC Course

PV Solar Plant Modeling for

Transient Stability Analysis
March/April 2021
IIT D and IIT DH

Vijay Vittal
Regents’ Professor
Ira A. Fulton Chair Professor
Arizona State University
Modeling Solar Photovoltaic Plants

2
 Generic model of a photovoltaic (PV)
solar power plant
• A generic model of a PV solar power plant is shown
on the next slide
• The key components of any generic model are the
inverter controls and the inverter model
• Typically, the solar power model is a user input and
is a function of the geographical location of the
plant

3
Modeling solar photovoltaic plants in stability studies
V ref bus
Vterm

Trip Signal
Ip (P)
Command

Control Converter
Model Model

Pgen, Qgen
IQ (Q)
Command

Power
Order

Solar Power
Model
(User-written)

4
 Generic model of a photovoltaic (PV)
solar power plant
• The converter is the main interface between the source of
power and the network
• Since a generic converter model has already been
developed for wind turbine generators, most transient
stability packages use this model for solar power plants also
• The converter is represented as a current injection device
which injects the required current into the network
• The active and reactive current commands are issued from
the control block
• The model receives the individual current commands and
injects a complex current into the network

5
 Converter Model
Rate limits on reactive current for recovery after fault.
Upper limit is active when Qgen0 > 0
Lower limit is active when Qgen0 < 0

Iqrmax
Iqcmd Iq High Voltage
1
1  sTg Reactive Current
Management

Interface to Network
Iqrmin Vt

Model
LVPL & rrpwr

Ipcmd 1 Ip
1  sTg
Low Voltage
LVPL
Reactive Current
Lvplsw

0
Management
LVPL1
1
1  sTfltr
1
Zerox Brkpt V

6
 Converter Model
• The time constants in the converter model represent
the switching of the solid-state switches within the
converter
• The behavior of a few limiting devices is also built
into this block
• The low voltage power logic (LVPL) uses the
terminal voltage magnitude to control the upper
limit on the active power injected
• If the bus voltage magnitude of the converter falls
below a certain threshold, due to the occurrence of a
disturbance, the LVPL block will reduce the upper
limit as per the characteristic shown
7
 Converter Model

• Within a range of low voltage magnitude values, the

active current upper limit is varied in a linear
fashion
• If the voltage magnitude falls below the lower
bound of the range, the active current upper limit is
made zero
• In the normal operating voltage range, the LVPL
block does not affect the active current upper limit
• Typically, all settings in the LVPL block can be set
by the user

8
 Converter Model

• Two algebraic current limiters are also present in

this block
• The high voltage reactive current management
section reduces the reactive power injected if the
terminal voltage magnitude rises above a certain
user defined limit
• The rate at which the reactive power is ramped
down in this case can be set by the user
• The low voltage active current management section
reduces the active power injected if the voltage
magnitude falls below a certain threshold and
functions similar to the LVPL block
9
 State Transition – switch position
State 0 – If Voltage_dip = 0, norm al operation (Iqing = 0)

Converter control model State 1 – If Voltage_dip = 1, Iqing goes to position 1

State 2 – If Thld > 0, then a fter the Voltage_dip goes ba ck to
zero, set va lue of I qfrz for t = Thld, after whic h go
back to state 0
If Thld < 0, then a fter the Voltage_dip re turns to zero
stay in State 1 for t = Thold, after whic h go back to
dbd1, dbd2 state 0

If (Vt < Vdip) or (Vt > Vup) then 1 - Verr Iqv

Voltage_dip = 1 Vt K qv
1  sTrv Vt _ filt =

Iql1
else +
Voltage_dip = 0 s0

Iqh1
Vref0 (user defined) 1
Freeze State if Voltage_dip =1 Iqfrz 2 0
pfref
tan Vmax Vmax Iqmax Iqmax
Q max I qinj
VFlag Q Flag +
Pe 1 + K qi 1 + K vi 1 I qcmd
1
1  sTp
  K qp +
s
K vp +
s +
- s3
s1 s2
Qext Iqmin 0
0 Q min Vmin 0 V Iqmin
(Qext is initialized to a Qgen min
Vt _ filt
constant, or can be pf Flag
connected to an external (s0)
Vref1 (user defined)
model) +
+
1
Iqinj = 0  1  sTiq VDL1
After 0 Current
Thld s4 PQFlag

0.01
seconds Limit

0 – Q priority
go to
State 0
If Voltage_dip = 1
 VDL2 Logic 1 – P priority

Vt _ filt
(s0)

If Voltage_dip = 0 Freeze State if Voltage_dip =1

Iqinj = Iqv ωg
2 1 Vt _ filt

0.01
Iqinj = Iqfrz Pmax I pmax Thld2
If Thld > 0 &
dPmax PFlag
(Pref is initialized to
Voltage_dip = 0 go to s6 1 1 Pord
State 2 If Thld , 0 &
Voltage_dip = 0 stay
a constant, or can
be connected to an
Pref   1+sTpord

I pcmd
in State 1 for Thld external model) 10
seconds s5
dPmin 0 Pmin
I pmin
 Converter control model
• The control model is responsible for generating the
active and reactive current commands for the
converter model
• The bottom section of the model shows the
calculation of the active current command. The
reference active power can be either set by the user
using an external user-written dynamic model or the
value scheduled in the power flow
• This features provides the flexibility for the
inclusion of a governor type model to set the active
power
• This is the WECC REGC-A model

11
 Converter control model

• The model also allows for three different ways of

setting the reactive current command:
1. PV Var controller emulator. This emulator is essentially
a voltage regulator. The terminal voltage magnitude is
compared to its reference value and the desired amount
of reactive power is calculated using a PI controller
2. Power factor regulator. The converter can be operated
at a desired power factor and the reactive power is
calculated based on the required power factor and the
active power
3. User defined reactive power. The user has an option to
provide the value of the required reactive power
• The converter current limit block ensures that all
current commands are within certain limits
12
 Model Validation Using Measurement
• The model described in the previous slides has been
verified for primary frequency response tests on two
actual PV plants
• The details can be found in the reference below
• This testing was done on a strong network and
hence the constant current source model worked
well

P. Pourbeik, S. Soni, A. Gaikwad, V. Chadliev, “Providing Primary

Frequency Response from Photovoltaic Power Plants,” CIGRE
Symposium, Dublin, Ireland, 2017.
13
14
15
16
17
18
 User defined converter control model
• The control model described previously works as
per design and specification but is quite complex
• In systems where there is significant penetration of
PV resources there is a need for a simpler control
structure bearing in mind that the stability of the
system hinges upon the interaction of these controls
with one another
• This model is a constant current source model and is
the WECC REGC-A model
• With this objective recent work at ASU has led to
the control design which will be discussed next

19
 References

• Ramasubramanian, D., Z. Yu, R. Ayyanar, V. Vittal,

and J. Undrill, “Converter Model for Representing
Converter Interfaced Generation in Large Scale
Grid Simulations,” IEEE Transactions on Power
Systems, Vol. 32, No. 1, pp. 765-773, January 2017.
• Ramasubramanian, D., V. Vittal, J. Undrill,
“Transient Stability Analysis of an all Converter
Interfaced Generation WECC System,” Paper #44
PSCC 2016, Genoa, Italy, June 20-24, 2016.

20
User defined converter control model

21
 User defined converter control model
• The effective real power order (Fig. b) on slide 21)
is a combination of the power set point and the
active power droop coefficient
• The reactive power order (Fig. a) on slide 21) is
obtained from the voltage error along with a
reactive power droop
• The QV droop is instrumental in obtaining a stable
operation between converters when multiple
converters are connected to the same bus
• The active power droop coefficient is denoted as Rp
and the reactive power droop coefficient is denoted
as Rq
22
 Equations describing the model

ds1
 K i Vref  s2  Rq Qactual  (1)
dt
ds2
 1 Tr   Vt  s2  (2)
dt
ds3
 1 TGpv    Pref    R p   s3  (3)
dt
Pcmd  s3 (4)
ds4
 K ip  Pcmd  Pactual  (5)
dt
ds5
 K iq Qcmd  Qactual  (6)
dt
Qcmd  s1  K p Vref  s2  RqQactual  (7)
I Qcmd  Qcmd Vt  s5 (8)
I Pcmd  Pcmd Vt  s4 (9)
23
 Modeling details

• Limits have been imposed on the maximum active and

reactive power and minimum reactive power deliverable
• In choosing the limits for the reactive and active power, it
has been assumed that the converter can withstand an
instantaneous MVA of 1.7 times its rating
• Further, it has been arbitrarily assumed that at a terminal
voltage level of 0.75pu, the minimum operable power factor
is 0.4
• As the voltage dips, the limits of the converter control will
change to allow for more reactive power to be delivered
while curtailing the active power delivered to meet the MVA
rating

24
 Modeling details

• Though a terminal voltage of 0.75pu has been chosen as the

minimum voltage, the maximum deliverable reactive power
is maintained constant for voltages below 0.8pu

Variation of Qlimit with Vt

25
 Modeling details
• The value of qmax1 is obtained from the power flow but is
assumed to be the value of maximum reactive power at a voltage
level of 1.0 pu
• The value of qmax2 is obtained as given by (10) below
1.7  MVA
2

qmax 2 
 
1
1   (10)

 tan cos 1 0.4
 
2

 

• Hence, at any voltage level Vt above 0.8 pu, the value of qmax is
obtained as given 2 11
qmaxby below
qmax1
qmax  qmax1  Vt  1.0  (11)
0.8  1.0
• The value of qmin is maintained constant as specified in the power
flow while the maximum active power is obtained as in (12) to
maintain the MVA rating
1.7 * MVA
2
pmax   qmax
2
(12)

26
 Converter model

• The converter is modeled as a controlled voltage source to

incorporate the effect of the converter coupling inductance
• As the converter is practically a controlled ac voltage source
with a voltage source on the dc side, the converter is
modeled as a Thévenin source as shown in the figure on the
next slide and given by (13).

Ed  Vtd 0  id R f  iq X f
Eq  Vtq 0  iq R f  id X f (13)

27
Converter model

28
 Converter model

• The impedance Rf + jXf represents the coupling inductance which

can either be a filter inductor or a transformer
• E and  represent the magnitude and angle of the developed
convertor voltage
• Vtd0 andVtq0 represent the dq axis components of the terminal
voltage Vt
• The current commands IQcmd and IPcmd represent the active and
reactive current commands provided by the outer loop control
• The time constants TQ and TD represent the effect of the inner
current control loop of the converter control structure
• The converter representation includes the effect of the dc voltage
and amplitude modulation ratio m of the pulse-width modulation
control depicted by the PWM block in the figure on the previous
slide
29
 Converter model

• To achieve a steady state modulation ratio of 0.6 the

carrier voltage (VT) and dc voltage (Vdc) are initialized to
be:
Ed2  Eq2 Ed2  Eq2
VT  ; Vdc  (14)
0.6 0.5  0.6

At every time step, the values of Ed and Eq from

(13) are used as follows:
• The magnitude and angleE of the required internal
 E E
voltage
 tan 1  Eq Ed 
2 2

is obtained as and d q

m  E VT
• The modulation index is calculated as . The value
of m is limited to be between 0.4 and 1.0.

30
 Converter model

Generation of switching pulses for an inverter

31
 Converter model
• The figure on the previous slide shows the generation of
switching pulses for the inverter and the relation between the
carrier voltage and the modulating/reference voltage
• While the proposed positive sequence converter model does
not represent the actual switching model of the inverter, the
effect of the modulation index on the ac output voltage has
been represented
• The phase voltage values are then obtained as
Ea ,b ,c  0.5mVdc cos s t    120i  where s =377 rad/s and i = 0,
1, 2
• The values of E*d and E*q (E* ) are obtained by applying
Park’s transformation on Ea, b, c

32
Simulation Result

• Loads treated as constant admittance

• Load at bus 6 was increased by 10
MW at t = 5s.
• The entire system was represented
with detailed models in PLECS (a
point on wave simulation software
• With PSLF where the converter had
two representation:
• Controlled voltage source
• Current sources

33
 Parameter Settings for Simulation
• The proportional and integral gains of the PI controller in the
reactive power loop were set as 4.0 and 20.0 respectively
• The integral gains Kip and Kiq were each 10.0
• The value of Rf = 0.0025 pu and Xf = 0.06 pu on the converter
MVA base
• The remaining controller parameters are as shown below

34
 Simulation Scenario

• Machine at bus 1 was replaced with a converter

• Machines at buses 2 and 3 were retained as
synchronous machines with associated governors
and static exciter models.

35
36
 Discussion of result

• The inset figure on the previous slide shows the response of the three
models at the instant of disturbance
• The controlled voltage source representation of the converter model and the
PLECS response capture the near instantaneous achievable by the converter
while the boundary current representation is unable to do so
• The difference in the peak value between the voltage source representation
response and the PLECS response can be attributed to the fact that the time
step of simulation is much smaller in PLECS which allows for the change in
the internally generated converter voltage upon recognition of the
disturbance
• In the positive sequence simulation however, at the instant of the
disturbance, the internally generated converter voltage is constant while the
terminal voltage changes
• Other dissimilarities can be attributed to the difference between the point on
wave modeling in PLECS wherein a differential R+sL model is used in
PLECS for the filter inductor whereas in the positive sequence phasor model
the filter is represented by its algebraic fundamental frequency resistance
and reactance in the Thévenin impedance
37
38
 Discussion of results
• The instantaneous three-phase voltage and current
waves at the converter terminals are shown on the
previous slide
• These figures show that there is negligible change in
the terminal voltage while the converter current rises
near instantaneously to meet the increase in demand
• The magnitude of the converter current output from
PLECS is compared with the positive sequence
simulations on the next slide
• The near instantaneous response achievable by the
converter is captured by the voltage source
representation of the converter
39
• The reactive power response is shown above. It can be immediately observed that the
voltage source representation response is the more acceptable positive sequence phasor
approximation to the point on wave simulation. From the inset of the figure it can be seen
that the reactive power trajectory of the boundary current simulation is inconsistent with
the result from the PLECS simulation.

40
41
 WECC Modeling WG Group
• The model depicted in the earlier slides became the
WECC WG model REGC-B for PV solar plants
• This model was improved by including the inner
current control loop in the reference below

42
 WECC Modeling WG

• In recent work, the REGC-B model with the inner

current control loop has been further improved to
included a phase locked loop PLL
• This is now referred to as the REGC-C model.
• This model is described in the reference below

D. Ramasubramanian, W. Wang, P. Pourbeik, E. Farantatos, A. Gaikwad, S. Soni,

and V. Chadliev, “Positive sequence voltage source converter mathematical model
for use in low short circuit systems,” IET Gener. Transm. Distrib., Vol.14, No. 1,
pp. 97-97, 2020

43
44
Flowchart for Algebraic Manipulation of Norton
Current

45
Positive Sequence Implementation of PLL

46
 Comparison of REGC-C Model with EMT
Simulation
• In the reference where the REGC-C model was
developed, the positive sequence model was tested
versus detailed simulation in an EMT software on a
weak synthetic network
• The model performed well and captured the
essential features of the PV resource
• There are no measurements in this case, but the
positive-sequence model results are comparted with
the EMT results with a detailed inverter model

47
48
49
50
Description of Legend for Figures on the Previous Two Slides

51
52
53
 Testing on a Portion of a Large Utility
System
• A portion of the network in a real transmission
footprint, with the addition of several hypothetical
renewable energy resources that are in the
interconnection queue is considered
• The system has 119 buses, 3 equivalent sources
(representing external system outside the studied
area) and 9 renewable power plants
• All the renewable power plants are modeled as Type
4 WTGs in voltage control mode with reactive
power priority during low voltage ride through
conditions

54
 Testing on a Portion of a Large Utility System

• The total renewable generation is 665 MW (total

capacity 894 MVA) with 393 MW transferred to the
equivalent sources
• In positive sequence, all nine renewable energy
sources were modeled with the proposed modified
converter model and the REEC_A electrical control
model in local voltage control mode
• The proportional and integral controller gains for
the voltage controller were set as 0.0 and 120.0,
respectively

55
 Testing on a Portion of a Large Utility System

• In the proposed modified converter model, re =

0.0015 pu, xe = 0.15 pu, Imax = 1.1 pu, KIp = 6.0, KIi
= 70.0, KPLLp = 60.0 and KPLLi = 1400.0
• The limits of the PLL (Δωmax and Δωmin) were set at
12 Hz
• For one particular wind turbine, WTG 5, the
frequency and angle from the PLL from both the
positive sequence simulation and the point on wave
simulation when the fault is cleared 1/8th cycle
before the critical clearing time, and when the fault
is cleared 1/8th cycle after the critical clearing time
56
Frequency and angle of PLL of WTG 5 from positive sequence and point on
wave simulation when fault is cleared 1/8th cycle before critical clearing time

57
Frequency and angle of PLL of WTG 5 from positive sequence and point on
wave simulation when fault is cleared 1/8th cycle after critical clearing time

58
 Testing on a Portion of a Large Utility
System
• It can be seen that the fast control loops in the
modified positive sequence model are able to
represent the transient behavior seen from the
detailed point on wave simulations for both the
stable and

59
Active, reactive power and terminal voltage magnitude of WTG 5 from positive
sequence and point on wave simulation when fault is cleared 1/8th cycle before
critical clearing time

60
Active, reactive power and terminal voltage magnitude of WTG 5 from positive
sequence and point on wave simulation when fault is cleared 1/8th cycle after
critical clearing time

61