Modeling of GE Solar

Photovoltaic Plants for Grid

Studies

Prepared by:
Kara Clark
Nicholas W. Miller
Reigh Walling

Version 1.1

April 16, 2010

General Electric International, Inc.

One River Road
Schenectady, NY 12345
USA
Legal Notice

This report was prepared by General Electric International, Inc. (GEII) as an account
of work sponsored by GE Energy business. Neither GE Energy nor GEII, nor any person
acting on behalf of either:
1. Makes any warranty or representation, expressed or implied, with respect to the
use of any information contained in this report, or that the use of any information,
apparatus, method, or process disclosed in the report may not infringe privately
owned rights.
2. Assumes any liabilities with respect to the use of or for damage resulting from the
use of any information, apparatus, method, or process disclosed in this report.
Copyright©2009 GE Energy. All rights reserved .

ii
Foreword

This document was prepared by GE Energy in Schenectady, NY. Technical and

commercial questions and any correspondence concerning this document should be
referred to:

Kara Clark
GE Energy
Building 53-302A, One River Road
Schenectady, New York 12345
Phone: (518) 385-5395
E-mail: kara.clark@ge.com

iii
Revision History
Version 1.0 – September 9, 2009
Version 1.1 – April 16, 2010

Summary of Changes in Version 1.1

• Updated inverter size from 600 kW to 700 kW
• Updated current limits in LVPL and electrical control models
• Updated reactive power control figure, added new data to table
• Updated electrical control figure, added text describing input test signal

iv
Table of Contents
1 INTRODUCTION.....................................................................................................................1.1

2 MODEL OVERVIEW ...............................................................................................................2.1

2.1 LOAD FLOW MODEL ..........................................................................................................2.2
2.2 INITIAL CONDITIONS FOR DYNAMIC SIMULATION ..................................................................2.4
3 DYNAMIC MODEL DESCRIPTION........................................................................................3.1
3.1 CONVERTER MODEL .........................................................................................................3.1
3.1.1 Fault Ride Through ....................................................................................................3.2
3.2 CONTROL MODEL .............................................................................................................3.5
3.2.1 Reactive Power Control .............................................................................................3.6
3.2.2 Electrical Control ........................................................................................................3.8
3.2.3 Control Strategies ....................................................................................................3.12
4 BENCHMARK SIMULATIONS...............................................................................................4.1
4.1 TEST SYSTEM ...................................................................................................................4.1
4.2 SIMULATION RESULTS .......................................................................................................4.2
4.2.1 Fault Response ..........................................................................................................4.2
4.2.2 Capacitor Switching Response ..................................................................................4.7
4.2.3 Zero-Power Operation................................................................................................4.9
4.3 SIMULATION DATA...........................................................................................................4.11
5 CONCLUSIONS......................................................................................................................5.1

v
Table of Figures
Figure 2-1. GE Solar Plant Dynamic Model Connectivity. ..................................................2.2
Figure 2-2. Simplified Solar Plant Power Flow Model Examples.........................................2.3
Figure 3-1. Converter Model. ..............................................................................................3.2
Figure 3-2. ZVRT Model Settings and Equipment Specification. ........................................3.4
Figure 3-3. Overall Reactive Power and Electrical Control Model. .....................................3.5
Figure 3-4. Reactive Power Control Model. ........................................................................3.6
Figure 3-5. Q Droop Function Model. .................................................................................3.8
Figure 3-6. Electrical Control Model. ................................................................................3.10
Figure 3-7. Converter Current Limit Model. ....................................................................3.11
Figure 4-1. Test System. ......................................................................................................4.1
Figure 4-2. Solar Plant (Nominal Reactive Capability) Response to Line Fault. ..................4.4
Figure 4-3. Solar Plant (Nominal Reactive Capability) Response to Prolonged Low Voltage
Event. ...........................................................................................................................4.5
Figure 4-4. Solar Plant (Extended Reactive Capability) Response to Prolonged Low Voltage
Event. ...........................................................................................................................4.6
Figure 4-5. Solar Plant (Extended Reactive Capability) Response to Shunt Capacitor
Switching. ....................................................................................................................4.8
Figure 4-6. Solar Plant (Extended Reactive Capability) Operation at Zero Power Output.
...................................................................................................................................4.10

Table of Tables
Table 2-1. Single Converter Rating. ....................................................................................2.3
Table 3-1. Zero-Voltage Ride Through (ZVRT) Thresholds and Durations. .......................3.3
Table 3-2. Converter Model Parameters. ............................................................................3.4
Table 3-3. Reactive Power Control Parameters (on Generator MVA Base). .......................3.7
Table 3-4. Q Droop Function Parameters............................................................................3.8
Table 3-5. Electrical Control Parameters. .........................................................................3.11
Table 4-1. Solar Plant Converter Model Data for Simulations. .........................................4.11
Table 4-2. Solar Plant Reactive Power and Electrical Control Model Data. ......................4.12

vi
1 Introduction
GE Energy is developing models of GE solar photovoltaic (PV) plants suitable for use
in system impact studies. This report documents the current recommendations for
dynamic modeling of solar plants that use photovoltaic cell arrays with a full converter
interface to the power system. Throughout this document, such a PV system is simply
referred to as a solar plant. Other types of solar generation, e.g. solar thermal facilities
that use concentrators and steam generators without a full converter interface, should
not be represented with this model. The model structure is based upon the GE full
converter wind turbine generator models.
The model provided is as detailed as is appropriate for bulk power system studies.
It is valuable to put the model limitations in the context of what analysis is required.
Most important, this model is for positive sequence phasor time-domain simulations –
e.g. PSLF or PSS/e. Second, this assumes that the analysis is mainly focused on how
the WTGs react to grid disturbances, e.g. faults, on the transmission system. Details of
the device dynamics have been substantially simplified. Specifically, the very fast
dynamics associated with the control of the converter have been modeled as
algebraic (i.e. instantaneous) approximations of their response. The model is not
intended for use in short circuit studies or electromagnetic transient studies.
This model represents a solar plant with a dedicated feeder to the grid
interconnection. It should not be used to represent distributed generation that is in
compliance with IEEE 1547 and UL 1741.

GE Energy 1.1 GE Solar Modeling-v1-1.doc, 4/16/10

2 Model Overview
A GE solar plant has many similarities to a wind plant that uses full converter wind
turbine generators. Both consist of multiple small sources of electrical power, which
are aggregated and injected into the transmission system at a single point. Both use a
converter interface to the power grid. Both must meet system performance criteria,
such as voltage regulation, reactive power control, and under-voltage tripping.
Therefore, the solar plant model described in this document is based upon GE’s wind
plant models as described in “Modeling of GE Wind Turbine-Generators for Grid
Studies“. As solar plant technology, control philosophy, and/or interconnection
requirements evolve, so will the model.
In practice, a solar plant has a local grid collecting the output from the converters
into a single point of connection to the grid. Since the solar plant is made up of many
identical converters, it is a reasonable approximation to parallel all the converters into
a single equivalent large converter. This approach is used for the models presented in
this report. However, there are limitations. Disturbances within the local collector grid
cannot be analyzed. A single converter equivalent requires the approximation that the
power output of all the converters will be the same at a given instant of time. For
system impact studies, simulations are typically performed with the plant at maximum
power output. Under this condition, the assumption that all converters are at the
same, rated, output is not an approximation. For other conditions, this assumption
presumes minimal geographic dispersion and uniform solar irradiation across the
plant. It is also suitable when the aggregate dynamic behavior of converters with
dissimilar array irradiations is about the same as that of an equivalent converter
whose array receives the average level of irradiation. Simulations of bulk system
dynamics using a single converter equivalent is adequate for most planning studies.
From a load flow perspective, standard generator and transformer models are
required for initialization of the dynamic simulation program.
The fundamental frequency electrical dynamic performance of a solar plant is
completely dominated by the converter. The control of active and reactive power is
handled by fast, high bandwidth regulators within the converter controls, and can be
greatly simplified for simulation of bulk power system dynamic performance. Two
device models, a converter model and an electrical control model, are used to
construct a solar plant model. An overview of this structure is shown in Figure 2-1.
The converter model injects real and reactive current into the network in response
to control commands, and represents low and high voltage protective functions (e.g.,
low voltage ride through capability). The real power signals are initialized to the
generator output in the power flow.
The control model includes closed loop reactive power controls, and voltage
regulation with either a simplified emulator of GE’s SolarCONTROL 1 solar plant

1 Trademark of General Electric Company

GE Energy 2.1 GE Solar Modeling-v1-1.doc, 4/16/10

supervisory control system or a separate, detailed control model. The control model
sends a reactive command to the converter model.
In addition, user-written models can inject power profiles into the solar plant
dynamic models, or to represent additional protective functions (e.g., over/under
frequency).
The model allows reasonable customization of the control parameters to meet
specific application requirements.
Vreg bus
Vterm

Trip Signal

Ip (P)
Command
Control Converter
Model Model
Pgen , Qgen
IQ (Q)
Command
Power
Order

Solar Power
Model
(User-written)

Figure 2-1. GE Solar Plant Dynamic Model Connectivity.

2.1 Load Flow Model

A simplified model of a GE solar plant is appropriate for load flow analysis. Such a
model is shown in Figure 2-2. This model consists of a single generator and unit
transformer with MVA ratings equal to N times the individual device ratings, where N is
the number of converters in the plant. This is sufficient for smaller solar plants with a
point of interconnection (POI) at 34.5 kV or below. Larger plants, connecting into higher
voltage substations, typically require a second transformer to step from the collector
voltage to the sub-transmission or transmission voltage level. Given the relatively
small footprint of a solar plant, the collector system impedance and charging are
neglected.
The aggregate solar plant is modeled as a conventional generator connected to a
480 V bus. The generator real power output (Pgen), maximum reactive power output
(Qmax), and minimum reactive power output (Qmin) should match the solar plant
capability. The reactive power capability will depend upon the number of converters
provided, and how that capability is used to meet the interconnection requirements.

GE Energy 2.2 GE Solar Modeling-v1-1.doc, 4/16/10

The capability of a single converter is shown in Table 2-1. The converter will prioritize
reactive power production over real power production to ensure the overall converter
current limits are respected.
To illustrate the trade-offs, consider a solar plant with a maximum real power
output of 10.5 MW. It could be configured with 16 converters for a total MVA rating of
11.3 MVA, which would provide a +/-0.93 power factor range (+/-4.2 MVAr) at full power
output. Alternatively, the same plant could be configured with 15 converters for a total
rating of 10.6 MVA. This second plant would limit the maximum power output to
9.7 MW under conditions requiring reactive power output at +/-4.2 MVAr. Or it could
deliver the full 10.5 MW, if the interconnection required no more than +/-0.99 power
factor (+/-1.45MVAr). The load flow model should accurately reflect the capabilities of
the plant under study.
A typical distribution substation interconnection would be at a 15, 25 or 35 kV-class
voltage level, requiring a suitably rated padmount transformer with an impedance of
about 6% on the transformer MVA rating. For larger plants, a substation transformer
would also be necessary. It should also be suitably rated for the size of the plant, with
an impedance of about 10% on the transformer MVA rating.

34.5 kV
POI Bus
Pgen
Qgen

138 kV
POI Bus
34.5 kV
Pgen Collection Bus
Qgen

Figure 2-2. Simplified Solar Plant Power Flow Model Examples.

Table 2-1. Single Converter Rating.

Generator Rating 707 kVA
Pmax 700 kW
Pmin 0 kW
Qmax 99 kVAr*
Qmin -99 kVAr*
Terminal Voltage 480 V
*These values are for +/- 0.99 power factor.

GE Energy 2.3 GE Solar Modeling-v1-1.doc, 4/16/10

2.2 Initial Conditions for Dynamic Simulation
The load flow provides initial conditions for dynamic simulations. The conditions
outlined above are generally applicable to the dynamic model presented below. The
maximum and minimum active and reactive power limits must be respected in order
to achieve a successful initialization.
If the solar plant’s electrical control is customized to meet a particular set of
desired performance objectives, then the load flow must be initialized in accordance
with those customized rules. For example, it is possible to inject or absorb reactive
power (e.g., regulate voltage) at zero real power with a converter system (SolarFREE
Reactive Power 2 ). Therefore, the real power at the generator in the power flow must
be zero for this type of simulation.
The dynamic solar plant models have the option to use the SolarCONTROL
supervisory control system to regulate a measured bus voltage, such as the POI. Line
drop compensation may also be used to regulate the voltage some distance into the
transmission system. If these features are selected in the dynamic models, then the
generator in the load flow model needs to regulate the corresponding bus.
Inconsistencies between the power flow and the dynamic model will result in an
unacceptable initialization.

2 Trademark of General Electric Company

GE Energy 2.4 GE Solar Modeling-v1-1.doc, 4/16/10

3 Dynamic Model Description
This section presents the engineering assumptions, detailed structure, and data for
each of the component models necessary to represent a GE solar plant.

3.1 Converter Model

This model (gewtg in PSLF) is the equivalent of the converter, and provides the
interface between the solar plant and the network. It is an algebraic, controlled-
current source that computes the required injected current into the network in
response to the real and reactive current commands from the electrical control model.
This controlled-current source also incorporates the low voltage power logic and the
fast-acting converter controls that mitigate over-voltages by reducing reactive current
output.
The model is shown in Figure 3-1. The real and reactive current command signals
are developed in the electrical control model described in Section 3.2. The real current
command signal is initialized to match the generation power output from the power
flow. It remains constant unless a user-written model is used to provide a power
profile that varies with time. The low-pass filters on the incoming command signals
are simple approximations to the complex, fast electronic control system. This small
lag (0.02 seconds) provides a reasonable representation in the time frame of interest.
As with all positive sequence fundamental frequency analysis, sub-cycle behavior is
not meaningful.
The Low Voltage Power Logic (LVPL) reduces system stress during and immediately
following sustained faults by limiting the real current command with both a cap and a
ramp rate limit. Under normal operating conditions, the filtered terminal voltage is
above a user-specified breakpoint (brkpt) and there is no upper limit. When the voltage
falls below the breakpoint during a fault, a cap is calculated and applied. When the
voltage is below a user-specified zero-crossing point (zerox), the cap becomes zero.
The user-specified ramp rate limit (rrpwr) is key to the post-fault power recovery.
During this recovery period, the voltage will exceed the breakpoint and the cap is
removed. However, the real current command rate of increase will be restricted by the
ramp rate limit.
The actual converter controls include a phase-locked loop (PLL) to synchronize with
the system. However, the PLL dynamics are extremely fast relative to the PSLF time
frame, and under normal grid operating conditions result in effectively perfect tracking.
Under transient conditions of severe voltage depression and relatively high system
impedance, delivery of active current becomes limited. The fast control actions
effectively result in reduced active current delivery. This fast action is captured in the
model by a low voltage active current management function. This is a linear reduction
of active current injection for terminal voltages below 0.8 pu. This effect is modeled
within the network solution (i.e. without state variables), which is consistent with the
overall algebraic modeling of current injection by the converter model. The reactive
current delivery remains high under these transient conditions, providing voltage
support and short circuit strength.

GE Energy 3.1 GE Solar Modeling-v1-1.doc, 4/16/10

 The fast controls will also act to limit excess voltage on the terminals by
suppressing reactive current injection when the terminal voltage rises excessively. This
effect is modeled by a high voltage reactive current management function in the
network solution, which drives reactive current injection down to limit terminal voltage
to 120%. Reactive current injection is limited to the machine rating.

Isorc
IQcmd 1
-1 High Voltage
(efd) 1+ 0.02s
Reactive Current
From s0 Management
ewtgfc
LVPL & rrpwr
Low Voltage
IPcmd 1 IPlv Active Current
(ladifd) 1+ 0.02s Management
From s1
ewtgfc

LVPL
Vterm
1.22

LVPL V
1
1+ 0.02s
V s2
zerox brkpt
(0.40) (0.90)
Low Voltage Power Logic

Figure 3-1. Converter Model.

3.1.1 Fault Ride Through

The converter model also includes over/under voltage protective functions. GE
solar plants have “zero voltage ride through” (ZVRT) capability. Voltage ride through
requirements are defined such that a plant must not trip for events that are less severe
than the defined thresholds and time durations. Plants may tolerate more severe
events without tripping. Use of the model therefore does not ensure that the plant will
trip, only that it is allowed to do so. The thresholds and time durations for this
protection will vary significantly from one project to another as equipment designs are
modified to meet specific grid codes or interconnection agreements.
Recommendations for modeling the protection functions are as follows:
• For feasibility and reliability impact studies of future projects: An objective
of the study should be to establish the voltage and frequency excursions that
may occur. Therefore, either do not include the protection model or else set the
trip levels consistent with applicable grid codes for the project.
• For facility studies for projects in the design phase: Use trip settings
consistent with performance commitments. The results of the study should

GE Energy 3.2 GE Solar Modeling-v1-1.doc, 4/16/10

 indicate acceptable settings for the actual protective devices to satisfy system
requirements while providing adequate protection for the equipment.
• For studies involving in-service projects: Use the actual trip settings of the
protective equipment.
Table 3-1 gives voltage trip levels and durations for the ZVRT option available with
SolarRIDE-THRU 3 . It is important to note that the low voltage thresholds are a stepwise
fit to a curve that defines the equipment minimum performance specifications. Figure
3-2 shows this graphically; with the step-wise curve representing the trip points in the
table. The step-wise curve is conservative, in that it is always inside the specification.
As noted above, low voltage ride through requirements vary from application to
application. The tripping thresholds and durations should be chosen to appropriately
represent the application under study.
Any other desired protective functions (e.g., over/under frequency) would need to
be implemented with additional protective device or user written models.

Table 3-2 includes recommended settings for the converter model. The maximum
allowed ramp rate limit, rrpwr, is 20, and the minimum allowed is 3. The LVPL
breakpoint, brkpt, must be greater than or equal to 0.4, less than or equal to 1.0, and
greater than the zero-crossing, zerox.

Table 3-1. Zero-Voltage Ride Through (ZVRT) Thresholds and Durations.

V (%) ΔV (pu) Time (sec)
75 -0.25 1.9
50 -0.50 1.2
30 -0.70 0.7
15 -0.85 0.2
110 0.10 1.0
115 0.15 0.1

3 Trademark of General Electric Company

GE Energy 3.3 GE Solar Modeling-v1-1.doc, 4/16/10

 140
Voltage at Point of Interconnection

120

100
(Percent)

80
PSLF
60

40
GE ZVRT
20

0
-1.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0
200 ms
Time (seconds)

Figure 3-2. ZVRT Model Settings and Equipment Specification.

Table 3-2. Converter Model Parameters.

Parameter Name Recommended Values
lpp 0.8
dvtrp1 -0.25
dvtrp2 -0.50
dvtrp3 -0.70
dvtrp4 -0.85
dvtrp5 0.10
dvtrp6 0.15
dttrp1 1.9
dttrp2 1.2
dttrp3 0.7
dttrp4 0.2
dttrp5 1.0
dttrp6 0.1
fcflg 1
rrpwr 10.
brkpt 0.9
zerox 0.4

GE Energy 3.4 GE Solar Modeling-v1-1.doc, 4/16/10

3.2 Control Model
This control model (ewtgfc in PSLF) dictates the active power to be delivered to the
system based on the power flow initial conditions or a user-written solar power profile
(Pord). It dictates the reactive power to be delivered based on the supervisory VAr
controller output (Qord). Qord can either come from a separate model, or from the
SolarCONTROL voltage and reactive control emulator function included in the control
model. Qord can also be held constant or determined by a power factor regulator.
The model consists of the following control functions:
SolarCONTROL Emulator
Power Factor Regulator
Electrical Control
The overall block diagram for the Reactive Power Control and the Electrical Control
is shown in Figure 3-3. These controls are described in more detail in the following
sections.

Vref SolarCONTROL
Vreg Emulation

Option for user-written model

Qcmd
Qref

PFAref Power
Factor
Regulator

Reactive Power Control IQcmd

(to generator model)
Electrical
Qgen
Controller IPcmd
(to generator model)
Vterm

Pord
(from wind turbine model)
Pelec Pdbr
(to wind turbine model)
(from generator model)
P,Q Priority Flag

Figure 3-3. Overall Reactive Power and Electrical Control Model.

GE Energy 3.5 GE Solar Modeling-v1-1.doc, 4/16/10

3.2.1 Reactive Power Control
A more detailed representation of the Reactive Power Control is shown in Figure
3-4.
The SolarCONTROL emulator function represents a simplified equivalent of the
supervisory VAr controller portion of the entire solar plant management system
(SolarCONTROL). The function monitors a specified bus voltage and compares it to a
reference voltage. Three regulated bus options are available: the terminal bus, a user-
specified remote bus (e.g., the POI), or a synthesized point in the power system. The
latter bus is synthesized from local voltage and current measurements, and the
compensating reactance, Xc. The regulator itself is a PI controller. The time constant,
Tc, reflects the delays associated with cycle time, communication delays, and
additional filtering in the controls. The voltage measurement lag is represented by the
time constant Tr. Table 3-3 gives suggested settings for the SolarCONTROL emulator
model.
The other reactive power control method available is power factor control. It is
enabled by setting pfaflg to 1. The data associated with this mode are also shown in
Table 3-3. The appropriate flag and gain settings to represent various control
strategies are described in Section 0.
SolarCONTROL Vrfq
Emulator (vref)
Vermx Qmax
Kiv/s
Vreg + +
1 s4 1
1/fN Qord
1+ sTr - + Qwv 1+ sTc
Kpv
s3 - Vermn 1+ sTv Qmin s5

Vqd s2

From Q Droop
Function If yes, freeze 2
integrators
< Vfrz?

PFAref Qref
(vref) tan (vref) Qord from separate model
0
Pelec 1 1 (vref)
1+ sTp
x
s6 0 -1 Qmax
pfaflg
1
Qord Qcmd

varflg Qmin

Figure 3-4. Reactive Power Control Model.

GE Energy 3.6 GE Solar Modeling-v1-1.doc, 4/16/10

 Table 3-3. Reactive Power Control Parameters (on Generator MVA Base).
Parameter Name Recommended Value
Tr (sec) 0.02
Tv(sec) 0.05
fN 1.0
Tc(sec) 0.15
Kpv 18.
Kiv 5.
Qmax (pu) 0.14*
Qmin (pu) -0.14*
Tpwr (sec) 0.05
Xc (pu) 0.0
Vermn (pu) -0.1
Vermx (pu) 0.1
Vfrz (pu) 0.7
*Provides +/- 0.99 power factor at the terminals. Limits of +/-
0.436 pu reactive power would provide +/- 0.90 power factor
at the terminals, but the converter system would need to be
oversized relative to the real power rating.

The PI gains, Kpv and Kiv, are field adjustable to meet performance objectives and
may be adjusted in the model, if necessary. The values given in the table are rough
upper limits, based on GE simulation and experience. They should be suitable for
systems with a short circuit capacity of 5 or more times the solar plant rating. These
higher gains will give better voltage response to grid voltages disturbances. However,
higher gains result in increased risk of instability – not unlike the way AVR gains can
destabilize conventional synchronous machines. As a system weakens, the effective
closed-loop response gets faster. Thus, selection of higher gains for system
performance must be accompanied by analysis that assures stable operation under all
credible operating conditions – especially the minimum short circuit strength condition.
The parameter, fN, is used to represent wind plants with a reduced number of wind
turbines on-line. For solar plants, this parameter should be set to 1.
The Q Droop function, shown in Figure 3-5, is a relatively slow-acting function that
reduces the effective voltage reference (Vrfq-Vqd) as reactive power changes. This
improves coordination between multiple integral controllers regulating the same point
in the system. By default, the Q Droop function is disabled. It may be enabled by
setting the gain parameter, Kqd, to a non-zero value. Typical data are shown in Table
3-4. There are three options for the reactive power input to this function: reactive
power generated by the WTG, reactive power flow in a user-specified branch, or a
synthesized reactive power. The latter is the reactive power flow in the user-specified
branch plus a secondary term, Xqd*Im2, where Im is the magnitude of the current
flowing in that branch.

GE Energy 3.7 GE Solar Modeling-v1-1.doc, 4/16/10

 Q Input 1 Vqd
Kqd
1+ sTlpqd
s7

Figure 3-5. Q Droop Function Model.

Table 3-4. Q Droop Function Parameters.

Parameter Name Recommended Value
Tlpqd (sec) 5.0
Kqd 0.04
Xqd (pu) 0.0

3.2.2 Electrical Control

A more detailed representation of the Electrical Control is shown in Figure 3-6. This
model is a simplified representation of the converter control system.
The volt/var control monitors the generator reactive power, Qgen, and terminal
voltage, Vterm, to compute the reactive current commands IQcmd required to meet the
Qcmd from the Reactive Power Control. The Qcmd signal is compared to the reactive
power generated by the converter, and the resulting error is integrated with a gain of
Kqi, to generate a voltage reference, Vref. Thus, the reactive power command is
implemented via a slowly changing voltage reference. The subsequent voltage control
block is significantly faster. The voltage reference is compared to the actual terminal
voltage, and the resulting voltage error is multiplied by a gain and integrated to
compute the reactive current command Iqcmd. Thus, a drop in terminal voltage, e.g., in
response to a system fault, results in an immediate large voltage error and an
increased reactive current command.
As noted above, the power order (Pord) is initialized to match the generation power
output from the power flow. It remains constant unless a user-written model is used to
provide a power profile that varies with time. The real current command signal, IPcmd, is
developed from this power order and the terminal voltage.
The dynamic braking resistor (DBR) function is provided for GE’s full converter wind
plants, and is therefore in this model. This function will not be included in a solar plant.
The associated data are set to zero.
The details of the converter current limit are shown in Figure 3-7. The objective of
this function is to prevent the combination of the real and reactive currents from
exceeding converter capability. Depending upon the value of a user-specified flag,
pqflag, either real or reactive power has priority. It is expected that GE solar plants will
give reactive power priority over real power.
When reactive power has priority, the calculation of the limit on the reactive
current begins by determining the minimum of a hard reactive current limit, Iqhl, and a
voltage dependent limit, Iqmxv. The voltage dependent limit will be equal to the steady-

GE Energy 3.8 GE Solar Modeling-v1-1.doc, 4/16/10

state rating of the solar plant (as defined by the input parameter Qmax) at 1.0 pu
voltage and will linearly increase as voltage drops. The maximum voltage dependent
reactive current limit is 1.6 pu at zero voltage. The minimum of Iqhl and Iqmxv is
compared to a maximum temperature dependent converter current, ImaxTD. That
minimum is the maximum limit, Iqmx, applied to the reactive current order, IQcmd. The
minimum reactive current, Iqmn, is the negative of this maximum limit. The remaining
converter current capability, SQRT(ImaxTD2 – IQcmd2), becomes the maximum, Ipmx,
applied to the real current order, IPcmd. No minimum is applied to the real current
order. Reactive power priority is recommended, which is equivalent to the default
value of 0 for pqflag.
When real power has priority, the real current order, IPcmd, is limited to the
minimum of the maximum temperature dependent converter current, ImaxTD, and a
hard active current limit, Iphl. The calculation of the limit on the reactive current begins
by determining the minimum of a hard reactive current limit, Iqhl, and the voltage
dependent limit, Iqmxv, as described above. The minimum of Iqhl and Iqmxv is compared
to the remaining converter current capability, SQRT(ImaxTD2 - IPcmd2). That minimum is
the maximum (capacitive) limit, Iqmx, applied to the reactive current order, IQcmd. The
minimum (inductive) reactive current, Iqmn, is the negative of the maximum. No
minimum is applied to the real current order.
An auxiliary test signal can be injected into the terminal bus voltage regulator via
model[@index].sigval[0], as shown at the top of Figure 3-6. A user-written dynamic
model (epcmod) is needed to generate the desired signal. The index of the solar
converter model (@index) can be obtained using the model_index function.
Table 3-5 includes recommended settings for the electrical control model. The
converter current limit, ImaxTD, is a function of time and operation. However, it is
constant in this model (1.7 pu) and not user-specified.

GE Energy 3.9 GE Solar Modeling-v1-1.doc, 4/16/10

 Auxiliary Test Signal
Qgen ( model[@index].sigval[0] )
Vmax Iqmx
- + IQcmd
Qcmd + Vref
KQi / s KVi / s
+ (efd)
s0 - s1
to converter
Vterm Vmin Iqmn
model, gewtg

P,Q Priority Flag

Converter Current Limit

Ipmx
Pord Porx . IPcmd
.
(vsig) (ladifd)
from power flow to converter
initial conditions or model, gewtg
user-written model
Vterm

+ Pdbr
Pelec - Dynamic Braking Resistor
from converter (elimt)
model, gewtg to dead-end

Figure 3-6. Electrical Control Model.

GE Energy 3.10 GE Solar Modeling-v1-1.doc, 4/16/10

 P,Q Priority Flag
(pqflag)

0 1

Iqmn Iqmx Vt Iqmx Iqmn

Q Priority Iqmxv P Priority

1.6
qmax
Vt
-1 1.0 -1
Iqmxv
Iqhl Minimum

Minimum Minimum

IPcmd
ImaxTD ImaxTD2 - IPcmd2

IQcmd
ImaxTD2 - IQcmd2
Iphl

Minimum Minimum

Ipmx Ipmx

Figure 3-7. Converter Current Limit Model.

Table 3-5. Electrical Control Parameters.

Parameter Name Recommended Value
Kqi 0.1
Kvi 120
Vmax 1.10
Vmin 0.90
Iphl 1.24
Iqhl 1.25
pqflag 0=Q priority
EBST 0.
Kdbr 0.

GE Energy 3.11 GE Solar Modeling-v1-1.doc, 4/16/10

3.2.3 Control Strategies
A variety of control strategies can be represented, including voltage regulation of a
remote bus or constant power factor control, both with and without the supervisory
SolarCONTROL. The various strategies can be implemented by setting the varflg,
pfaflg, and Kqi parameters as follows:
• Operation with SolarCONTROL and with “volt/VAr” control (varflg = 1, pfaflg
= 0, Kqi = 0.1) This represents the normal configuration for a North American
solar plant, using the SolarCONTROL emulator in the model to represent the
plant level supervisory control.
• Operation without SolarCONTROL and with “volt/VAr” control (varflg = 0,
pfaflg = 0, Kqi = 0.001) With the SolarCONTROL turned off, Kqi is reduced so
there is a slow reset to desired reactive power and terminal voltage control is
rapid. This combination of flags and Kqi = 0.1 can be used to emulate
SolarCONTROL at a fixed plant reactive power control.
• Operation without SolarCONTROL and with fast power factor control (varflg
= 0, pfaflg = 1, Kqi = 0.5) This represents a configuration where a set power
factor angle is rapidly regulated by the converter control. Closed loop voltage
control is not used on these systems, but is left in the model to approximately
represent other means that are used to limit voltage excursions that would
otherwise cause unit tripping.
• Operation with SolarCONTROL and with fast power factor control (varflg = 1,
pfaflg = 0, Kqi = 0.5) This represents a reactive power control configuration
that includes SolarCONTROL. It is similar to the first control option except the
regulator gain is at a higher value. The power factor control flag, pfaflg, is set to
zero because the signal from the SolarCONTROL is a reactive power order,
rather than power factor angle.

GE Energy 3.12 GE Solar Modeling-v1-1.doc, 4/16/10

4 Benchmark Simulations
The models described in this report have been implemented in GE’s PSLF load flow
and dynamic simulation software. Representative results using the PSLF models are
presented in this section. Note that these simulations are not necessarily updated with
each version of this document. Therefore, the simulations may not always use the
latest model or data recommendations. They do, however, illustrate the general
performance characteristics of the solar plant model. The data used for these
simulations is shown in Section 4.3.
Upon request, GE will provide the PSLF benchmark simulation results to those who
wish to implement the models in other simulation programs. The results can be
supplied in ASCII format for cross-plotting in order to validate the model
implementation.
4.1 Test System
One line diagrams of the test system are shown in Figure 4-1. The top diagram
shows real and reactive power flow (MW, MVAr), and the bottom diagram shows
impedances (pu on 100 MVA). The test system represents an aggregate model of a
solar plant, with a single generator and substation transformer to the 34.5 kV point of
interconnection (POI). The solar plant is rated 10 MW.

Figure 4-1. Test System.

GE Energy 4.1 GE Solar Modeling-v1-1.doc, 4/16/10

4.2 Simulation Results
The following subsections describe simulation results illustrating the solar plant’s
overall performance, including voltage regulation, zero-power operation capability,
converter current limit, and low voltage power logic. A variety of disturbances are
represented – fault disturbances, capacitor switching events, and solar power profiles.
Plotted results are provided for each test simulation. Traces plotted together share
the same scaling, as shown on the y-axis. Scaling for a particular trace can be
confirmed by checking the legend below each plot. The same plot format was used for
all simulations, and a brief description follows.
From top to bottom, the left column shows the solar plant terminal voltage (dark
blue line, pu), point of interconnection voltage (red line, pu), and infinite bus voltage
(green line, pu) in the first plot. The second plot shows the solar plant’s real power
output (dark blue line, MW). The third plot shows the final power order from either the
power flow initialization or a user-written model to the electrical control model (dark
blue line, pu). The fourth plot shows the real power current command from the control
to the converter (ipcd, dark blue line, pu), the power command after the low voltage
power logic (iplv, red line, pu), and the maximum real power current limit (ipmx, green
line, pu).
From top to bottom, the right column shows the SolarCONTROL emulator voltage
reference (vrfq, dark blue line, pu) and regulated voltage (vreg, red line, pu). The second
plot shows the solar plant’s reactive power output (dark blue line, MVAr). The third plot
shows the reactive power order from the reactive power control (dark blue line, pu).
The fourth plot shows the reactive power current command (iqcd, dark blue line, pu),
the voltage dependent reactive current limit (iqxv, red line, pu), the maximum reactive
current (iqmx, upper green line, pu) and the minimum reactive current (iqmn, lower
green line, pu).
Note that per unit values of real and reactive power are on the MVA base of the
converter model.
4.2.1 Fault Response
Three simulations illustrate the solar plant’s response to grid disturbances. In all
cases, the regulated bus was the 34.5 kV POI and the SolarCONTROL emulator was
active. The converter current limits were implemented with Q priority.
In the first case, the 10 MW solar plant was configured with the minimum number
of converters for a 0.99 power factor at the terminal bus. This is equivalent to a rating
of 10.1 MVA, and a reactive capability of +/- 1.4 MVAr. The disturbance was a 3-phase
fault to ground cleared by tripping one of the 34.5 kV lines from the POI bus to the
Infinite bus. The solar plant’s response is shown in Figure 4-2. The reactive power
output of the solar plant was at its maximum during the fault in an effort to support
voltage. After the fault was removed, the SolarCONTROL emulator quickly restores the
voltage at the POI bus to its initial value. The effect of the low voltage power logic is
seen in the post-fault real power recovery over 250 msec.

GE Energy 4.2 GE Solar Modeling-v1-1.doc, 4/16/10

 In the second case, the 10 MW solar plant was still configured for a 0.99 power
factor. The disturbance was a prolonged, low voltage on the system. The solar plant’s
response is shown in Figure 4-3. In an effort to support the POI voltage, the solar plant
was supplying maximum reactive power. Due to the Q priority in the converter current
limiter, the real power output was reduced to below 9 MW for the duration of the low
voltage event.
In the third case, the disturbance was again a prolonged, low voltage on the
system. However, the 10 MW solar plant was configured with sufficient converters for
a 0.90 power factor at the terminal bus. This is equivalent to a rating of 11.1 MVA, and
a reactive capability of +/- 4.7 MVAr. Such a reactive capability could be required by
the host utility or ISO for large or transmission connected solar plants. The solar plant’s
response is shown in Figure 4-4. With the additional reactive capability, the solar plant
supplied both maximum reactive power and rated real power output during this low
voltage event.

GE Energy 4.3 GE Solar Modeling-v1-1.doc, 4/16/10

 Figure 4-2. Solar Plant (Nominal Reactive Capability) Response to Line Fault.

GE Energy 4.4 GE Solar Modeling-v1-1.doc, 4/16/10

 Figure 4-3. Solar Plant (Nominal Reactive Capability) Response to Prolonged Low Voltage
Event.

GE Energy 4.5 GE Solar Modeling-v1-1.doc, 4/16/10

Figure 4-4. Solar Plant (Extended Reactive Capability) Response to Prolonged Low Voltage
Event.

GE Energy 4.6 GE Solar Modeling-v1-1.doc, 4/16/10

4.2.2 Capacitor Switching Response
The solar plant’s response to a capacitor switching event was also tested. The
10 MW solar plant was configured with sufficient converters for a 0.90 power factor at
the terminal bus. As noted above, this is equivalent to a rating of 11.1 MVA, and a
reactive capability of +/- 4.7 MVAr. The regulated bus was the 34.5 kV POI and the
SolarCONTROL emulator was active. The converter current limits were implemented
with Q priority.
The solar plant’s response is shown in Figure 4-5. At 1 seconds, a 4.5 MVAr shunt
capacitor bank was switched in, and the POI and terminal bus voltages jumped by
about 3%. The volt/var function reacted to the increase in terminal bus voltage with
an immediate reduction of about 2 MVAr in reactive power output. Then the
supervisory SolarCONTROL emulator acted to further reduce the reactive power output
to its minimum. After about 10 seconds (more than is shown in the plot), the POI
voltage settles out to just slightly (~0.3%) above its starting value. The solar plant has
used all of its available reactive capability to return the POI bus voltage to its reference
level.

GE Energy 4.7 GE Solar Modeling-v1-1.doc, 4/16/10

 Figure 4-5. Solar Plant (Extended Reactive Capability) Response to Shunt Capacitor
Switching.

GE Energy 4.8 GE Solar Modeling-v1-1.doc, 4/16/10

4.2.3 Zero-Power Operation
An example of zero-power operation is shown in Figure 4-6. A user-written solar
power profile was used to emulate a decline in solar plant output as the sun sets. The
speed of this decline was set arbitrarily, with the objective of illustrating the zero-
power function. During the decline in power output, minor changes in reactive power
output were observed as the solar plant acted to regulate POI voltage. At about 105
seconds, the solar plant output reached zero. The solar plant, however, continued to
generate reactive power in order to regulate voltage at the POI bus. This is further
illustrated when a 4.5 MVAr shunt capacitor bank is switched in at about 110 seconds.
The solar plant reduced its reactive power output to its minimum, to return the POI bus
voltage at its initial value.

GE Energy 4.9 GE Solar Modeling-v1-1.doc, 4/16/10

 Figure 4-6. Solar Plant (Extended Reactive Capability) Operation at Zero Power Output.

GE Energy 4.10 GE Solar Modeling-v1-1.doc, 4/16/10

4.3 Simulation Data
The converter and electrical control dynamic data used in the benchmark simulations
are shown in Table 4-1 and Table 4-2, respectively.

Table 4-1. Solar Plant Converter Model Data for Simulations.

Model Parameter Value
MVA 10.1 or 11.1
lpp 0.80
dvtrp1 -0.25
dvtrp2 -0.50
dvtrp3 -0.70
dvtrp4 -0.85
dvtrp5 0.10
dvtrp6 0.15
dttrp1 1.90
dttrp2 1.20
dttrp3 0.70
dttrp4 0.20
dttrp5 1.00
dttrp6 0.10
fcflg 1
rrpwr 5.0
brkpt 0.90
zerox 0.50

GE Energy 4.11 GE Solar Modeling-v1-1.doc, 4/16/10

 Table 4-2. Solar Plant Reactive Power and Electrical Control Model Data.
Model Parameter Value
varflg 1
kqi 0.10
kvi 120.
vmax 1.10
vmin 0.90
qmax 0.14 or 0.43
qmin -0.14 or -0.43
tr 0.02
tc 0.15
kpv 18.
kiv 5.
pfaflg 0
fn 1.0
tv 0.05
tpwr 0.05
iphl 1.11
iqhl 1.25
pqflag 0
kdbr 0.
ebst 0.

GE Energy 4.12 GE Solar Modeling-v1-1.doc, 4/16/10

5 Conclusions
The solar plant model presented in this report is based on presently available
design information and engineering judgment. It is expected to give realistic and
correct results when used for bulk system performance studies. The modeling of solar
plants for bulk power system performance studies, however, will continue to evolve as
equipment improves, interconnection requirements change, and factory or field test
data becomes available.
This document is continuously being updated to reflect these changes. Those
using this document for modeling purposes are encouraged to verify that they are
using the most up-to-date version. This document is available through the PSLF
software website.

GE Energy 5.1 GE Solar Modeling-v1-1.doc, 4/16/10