1

Solar Dish-Stirling Power Plants and

Related Grid Interconnection Issues
Yixin Zhang, Senior Member, IEEE, and Bruce Osborn
Stirling Energy Systems, Inc.

underground electrical collection grid and transferred to

Abstract--A unique solar technology is entering the substations where the power can be transformed into extra-
competitive utility-scale power plant market. This solar dish- high voltage (EHV) for transmission across the utility grid.
Stirling technology combines a mirrored concentrator dish with a Figure 1 illustrates the basic operation of the system which
high-efficiency Stirling engine specially designed to convert can generate up to 25 kW of grid-quality electricity per
sunlight to electricity. In the next a few years, two large-scale system.
power plants will be constructed in California using this
innovative technology with a potential total capacity of 1,750
MW. The tens of thousands of small induction generators
installed in the power facility create a new set of challenges for
utility grid interconnection. The article introduces the solar dish-
Stirling technology and related interconnection issues including
power factor correction, self excitation over voltage, and low
voltage ride through. Options to meet the grid interconnection
requirements are also discussed.

Index Terms—AC generators, circuit breakers, modeling,

overvoltage protection, power conditioning, power system
stability, reactive power, solar energy, solar power generation,
static VAR compensators.

I. INTRODUCTION

A n overview is presented on the challenges of grid

interconnection with utility-scale solar power plants
using numerous small independently operating
Fig. 1 Solar dish-Stirling electrical power system

generators. Two large power plants are in the development stages for
In the past, the majority of solar generation was installed operation in California supplying up to 1,750 megawatts of
for small-scale remote-use application. For this application, power capability, representing the world’s two largest solar
photovoltaic ("PV") panels have been the best-suited power plants. Collectively, these power plants require up to
technology. As utilities gain more involvement with solar 70,000 solar dish-Stirling systems.
generation however, solar thermal technologies, especially Figure 2 presents an artist rendering of the utility-scale
concentrating solar power ("CSP") is gaining more attention. solar dish-Stirling power facility.
The dish-Stirling system is one of the CSP technologies. It
concentrates the solar energy using mirrors in a large dish.
The energy heats a working gas that drives a Stirling engine.
The engine then powers up an induction generator to provide
electricity to the grid. The system was initially developed in
the early 1980s and started field-testing in 1982. Significant
progress has been made over the last quarter century with
valuable operation knowledge transferred to the latest design.
The dish-Stirling system is one of the preferred solar
technologies because of the inherently high overall system
efficiency in converting sunlight to grid-quality electricity and
the attendant potential for lowest cost solar electricity. The
generator of each solar dish-Stirling unit is connected by an

Fig. 2 Utility-scale concentrating solar power facility.

1-4244-1298-6/07/$25.00 ©2007 IEEE.

 2

II. TECHNICAL CONSIDERATIONS AND CHALLENGES TABLE 1

REACTIVE POWER CALCULATION FOR A 25 kW INDUCTION GENERATOR
Unlike small distributed generation (DG) systems, large
capacity solar power plants need to meet strict kVAR kVAR
Rated Lagging kVAR
interconnection requirements to be a significant generating kW required for required for
PF Without PFC
0.95 leading 0.95 lagging
part of the grid. This involves power factor correction, over-
5.8 0.448 11.57 13.48 9.66
voltage protection, and low-voltage ride-through and
10.6 0.646 12.53 16.01 9.05
associated voltage recovery process. The following sections
15.9 0.739 14.5 19.73 9.27
discuss the implications of these requirements for large
22.4 0.804 16.57 23.93 9.21
(hundreds of MW capacity) dish engine solar plants.
25 0.815 17.78 25.99 9.56

A. Power Factor Correction (PFC)

Centralized solar power plants are required to provide Over the output power range of 5.8 kW to 25 kW requires
power factor correction capability (typically over the range of providing reactive power ranging from 9.05 kVAR to 25.99
0.95 lagging to 0.95 leading) at the Point of Interconnection if kVAR per dish to meet the 0.95 lagging to 0.95 leading
the Transmission Provider’s System Impact Study shows that requirement.
such a requirement is necessary to ensure safety or reliability. Providing 25.99 kVAR for a 900 MW (36,000 dishes) solar
Furthermore, the power factor correction equipment must power plant would require a total reactive power of:
provide a variable source of reactive power. The correction
can be either continuously variable or switched in discrete 25.99 kVAR × 36,000 = 936 MVAR
steps that are not allowed to change voltage level by more
than one percent per step. The capacitors must therefore provide a very large amount
There are two major groups of options to obtain the of reactive power.
reactive power needed for solar power plants. The first is Since the requirement is to provide the power factor at the
generator based and the second is capacitor based as Point of Interconnection, a small amount of VARs will need
discussed in the following sections. to be added for the reactive power loss from the transformers
in the two levels of substations. Also, cable capacitance in the
collection system should be included. Therefore, the final
B. Generator Based PFC Options
amount of the required reactive power generated by PFC
Many types of generators can dynamically produce capacitors will need to be adjusted accordingly.
reactive power with low associated operation costs. However, The data from the last column in Table 1 shows that to
for small generators, the major drawback of the generator reach the compensation level of 0.95 lagging, a 9.5 kVAR
based PFC options is a relatively higher initial equipment fixed capacitor in parallel with a 25 kW induction generator
cost. To use synchronous generators, DFIG (Doubly Feed would provide adequate reactive power at all power levels.
Induction Generators) or DC generators also requires Therefore, the adjustable part of reactive power for a 900
redesigning other parts of the systems. At this point, it is MW power plant will be:
assumed that induction generators will be used in the systems.
Synchronous condensers running without mechanical load 936 MVAR – 9.5 kVAR × 36,000 = 594 MVAR
can absorb or generate reactive power depending on the
excitation. When used with a voltage regulator, the condenser
can automatically run over-excited at times of high load and D. Mechanically Switched Capacitors (MSC)
under-excited at light loads. The disadvantage of the From an economic point of view, using MSC at
synchronous compensator is its high cost. The other possible intermediate voltage levels would normally be the most
drawback is when one of the centralized condensers needs common and preferred solution. By switching in and out a
repair or maintenance that may impact the operations of a portion or all of the capacitors in a capacitor bank, MSC can
large group of dishes, if the grid demands high power factor provide power factors in discrete steps. MSC typically have
during that time period. relative lower equipment cost comparing with other types of
switched PFC devices.
In this application, special attention should be given to the
C. Installed Reactive Power
lifespan of the mechanical switching devices. The lifespan of
Based on the PFC requirement, Table 1 shows the circuit breakers in the MSC is largely dependent on how
calculated reactive power data required for a single 25 KW frequently they are being switched. In typical power grid
induction generators. applications they may be switched on in the morning and off
in the evening, or twice a day. However, because solar power
generators have significantly varying output power levels
throughout a day, the reactive power levels would have to be
 3

adjusted accordingly and substantially several times each day. voltage is possible, if a reliable early detection can be
A high and fixed power factor requirement would result in obtained. The signals could come from an upstream
frequent switching and shorten the operational lifespan of the substation that switches the solar power plant into an island,
circuit breakers. Therefore, more study needs to be done on and/or from real-time three-phase instant voltage monitoring.
this type of circuit breaker selection, and the concomitant With an early warning signal and solid state contactors, the
operation and maintenance cost, including partial power plant generators may be switched off from the grid in time (which
output loss if one or more of these circuit breakers are out of disassembles the circuit that forms self excitation). If TSC is
service. used for the adjustable portion of PFC then it may quickly
The MSC devices could have additional problems by switch off the capacitor banks from the system and
producing switching inrush currents and associated over disassemble the circuit.
voltage on capacitors that could in turn shorten capacitor life. When a generator operating at full power is suddenly
The slow response time of MSCs would not help on over switched off the grid, it is important to prevent its speed from
voltage protection and regulation during transient or system getting too high, which could destroy the generator
recovery periods. mechanically. One possibility is to use a gas management
valve that short-circuits the high and low pressure sides of the
Stirling engine. This would significantly reduce the mechanic
E. Thyristor Switched Capacitors (TSC)
output and hence generator power to almost zero in a very
The TSC part of Static VAR Compensators (SVC) is short time period. More experiments need to be conducted to
another possibility. Since other portions of SVC, i.e. tuned verify whether or not the mechanic time delay in the engine
filters (FC) and thyristor controlled reactor (TCR) are not gas management system is short enough to be acceptable.
required, it should be less expensive than regular SVC. TSC Transient analysis also needs to be conducted to determine
can provide fast and smooth response without limits on the extent of over voltage, rise time and duration and the need
number and frequency of switching. This is ideal for dynamic for any special protective mechanisms to preserve all the
voltage regulation, flicker mitigation and power factor devices in the systems.
compensation. However, TSC equipment is more expensive
than MSC and has more energy loss during operation.
TSC’s fast response times may help to limit the over G. Low Voltage Ride-Through (LVRT) Capability
voltage as well (see Self Excitation section below). Low Although there are no technical standards related to LVRT
Voltage Ride-Through (LVRT) is another reason to select for solar power generating plants, they currently are expected
TSC, when large amounts of fast reactive power are needed to to meet the same regulation set by FERC for wind generating
help system voltage recovery (See LVRT section below). plants.[2] The FERC order states that “a plant shall be able to
remain online during voltage disturbances up to the time
periods and associated voltage levels set forth in the standard
F. Self Excitation and Over Voltage Risk
[in Figure 4 below] if the Transmission Provider’s System
A problem in reactive power planning that is easily Impact Study shows that the LVRT capability is required to
overlooked is the possibility of “self excitation”. This occurs ensure safety or reliability”. The key feature of this regulation
when the induction machine(s) become unexpectedly isolated is that a generating plant must have LVRT capability down to
from the grid with more capacitive current available than its 15 percent of the rated line voltage for up to 0.625 seconds.
no-load exciting current. In this application, the capacitive To meet this requirement and ensure the safety of the
current is not only generated by PFC capacitors but also by generators, the over speed must be prevented at the low
the capacitances from connecting transmission lines and output voltage for the generators. The short circuit valve
collection system cables. The LC resonance condition is met, mentioned above can be used to reduce the mechanic output
when the equivalent circuit has enough capacitance in parallel of a Stirling engine connected to the generator. Thus, over
with the generator inductance creating undesired over speed may be minimized ensuring the generator remains
voltages. Over-voltages greater than 10 times the rated voltage online.
can occur as a result of self excitation conditions. Without In addition to the FERC regulation, utilities also want solar
early detection and special protection mechanism power plants to provide additional capabilities to achieve
implemented, the only limits are saturation of the desirable voltage recovery and system stability after the 0.625
transformers, that may create destructive ferroresonance, or seconds. One is to maintain a proper and narrow band of
by an insulator flashover or a lightning arrester RPM range for the generators during the LVRT so that right
failure/explosion. A few wind power farms using induction after LVRT the frequency of the generators would be close to
generators have already experienced this type of severe grid frequency. The second is to provide controllable, fast
accident. response reactive power to meet the dynamic voltage stability
Self excitation is unavoidable in this application by using needs of the system. Maintaining RPM range may be achieved
induction generators and extremely high values of PFC by modulating the short circuit gas valve, if the Stirling engine
capacitance. However, protecting the systems from the over and the gas management valve can respond quickly enough.
 4

The dynamic system stability can only be met if TSC or other into mechanical motion. This power drives an induction
thyristor and capacitor based PFC devices are used for the generator, which produces electricity with an output of 3-
adjustable part of PFC capacitor banks. More steady state and phase, 480 Volts and 60 Hertz, so it is already power-
dynamic stability modeling and system level tests are conditioned by the generator interface.
necessary to determine the adequacy of these potential The Stirling engine incorporates a sealed 4-cylinder
solutions. reciprocating piston assembly along with gas coolers, gas
1.2 regenerators and heater heads. Concentrated solar energy
Voltage at the Point of Interconnection

1.1 heats the self-contained working gas (hydrogen) within the

Beginning of Emergency Low Voltage
1.0 heater head causing the gas to expand, pushing pistons
0.9 connected to crank shafts and an output shaft that drives a
0.8
(In per unit*)

Plant Required to
conventional electrical generator. There is a 90-degree timing
0.7 Remain On-Line separation between adjacent cylinders and the working gas is
0.6
alternately heated and cooled through repeated exchange
0.5
0.4 (back and forth) between the same adjacent cylinders. This
0.3 thermodynamic cycle is repeated over and over as the engine
0.2 Plant not Required to Remain On-Line
drives the induction generator to create electricity at a
0.1 frequency slightly higher than the grid.
- The Model Power Plant, and collocated indoor Power
-1.0 0.0 0.625 1.0 2.0 3.0 4.0 Conversion Unit test cells using a quartz lamp bank or a gas
* per unit = Ratio of Actual to
Time (seconds)
Nominal Voltage
fired burner as a heat source, permit dynamic testing under
both controlled and actual power plant operating conditions.
Fig. 3 Required Response to Emergency Voltage
This provides key data to correlate and update systems
models.
Following test bed validation, full scale testing on a 1MW
H. Test Beds for Grid Integration Power Group (40 solar dish-Stirling systems) is planned for a
The National Solar Thermal Test Facility (NSTTF) located Pilot Plant. These validation tests will be the precursor to a
at Sandia National Laboratories (SNL) will serve as the test staged ramp-up for the large-scale solar power facilities.
bed site for integration and validation testing. Figure 5
presents a photograph of the 150kW Model Power Plant
I. Summary
(MPP) comprised of six fully operational solar dish-Stirling
systems and located at the NSTTF. There are numerous challenges associated with integrating
new technologies for utility-scale applications. The many
benefits of solar dish-Stirling systems warrant additional
investigation to realize the potential of this highly efficient
renewable power technology. Near-term plans include further
modeling of the systems and subsystems to optimize the
performance and grid system interconnection while ensuring
the lowest possible life-cycle costs and related cost of energy.
Key areas of investigation and modeling include PFC, self
excitation and LVRT. Test bed testing of the most promising
solutions is currently planned for the Model Power Plant
located at the National Solar Thermal Test Facility. Following
Fig. 4 Model Power Plant Test Bed at the NSTTF test bed validation, full scale testing on a 1MW Power Group
will be performed at a Pilot Plant. These key stepping stones,
Each of the six systems is a 25-kilowatt electric solar systems modeling, test bed validation and pilot plant testing,
power unit designed to automatically track the sun and collect will ensure successful integration of utility-scale solar power
and focus solar energy onto a power conversion unit (PCU) plants onto the nation’s power grids.
that in turn converts the intense heat to grid-quality electricity.
The concentrator consists of a 38-foot-diameter dish
structure that supports 82 curved glass mirror facets, each III. ACKNOWLEDGEMENT
three-feet by four-feet in area. These mirrors concentrate and
focus the solar energy into the PCU receiver. The authors would like to express their sincere thanks to the
The PCU consists of the solar receiver, high-efficiency DOE for continued access to the NSTTF and the support and
Stirling engine, AC generator, cooling system, controls and contributions from the scientific and engineer teams at SNL
other support components. Power is generated by heat and Stirling Energy Systems.
transfer from the concentrated radiant solar energy to the
engine heater head and working gas that is finally converted
 5

IV. REFERENCES Group of L-3 Communications, Senior Director at Western

Periodicals: Digital Corporation, and started his career with Ford Motor
[1] R. C. Bansal, T.S. Bhatti, and D.P. Kothari, “Bibliography on the Company’s Aeronutronic Division. During his career, Mr.
Application of Induction Generators in Nonconventional Energy Systems,” Osborn has held diverse and key program, engineering,
IEEE Transactions on Energy Conversion, Vol. 18, No. 3, September
2003.
business development and operational management positions
[2] USA Federal Energy Regulatory Commission, “Interconnection for Wind with demonstrated performance in research, development,
Energy,” 18 CFR Part 35, Docket No. RM05-4-000 – Order No. 661, production programs, operations and high volume
June, 2005. manufacturing for domestic and international markets.
[3] C. S. Demoulias and P. S. Dokopoulos, ”Transient behaviour and self-
excitation of wind-driven induction generator after its disconnection from Mr. Osborn graduated with a B.S., Mechanical Engineering
power grid,” IEEE Trans. Energy Conversion, vol.5, pp.272-278, June from UC Irvine, a M.S., Systems Management from USC, and
1990. a Presidential/Key Executive MBA from Pepperdine
[4] Sharke, Paul. “Sun Rises on Solar.” Design News January 9, 2006: 58-62.
[5] Thompkins, Joshua. “Dishing Out Real Power.” Popular Science February University, all three with honors. He has U.S. Patents in solar
2005: p 31-33. energy and thermal systems, a NASA Award for technical
[6] Burroughs, Chris. “Utility-Scale Solar,” Sandia Technology Vol. 6, No. 4, innovation, and other scholastic and professional awards. He
Winter 2005: 12-14.
has published numerous peer-reviewed technical papers and is
an active member in several professional organizations.
V. BIOGRAPHIES

Yi Xin Zhang (SM’02) After

receiving his M.S.E.E. degree in 1981
from the Graduate School of
Electrical Power Research Institute,
China, Mr. Zhang worked with
Chinese and US National
Laboratories for 10 years on high
voltage test and measurements
technology.
Mr. Zhang has been working with
R&D and manufacturing companies
in the US during past 16 years. His special fields of interest
include HV test and measurements, electrical and optical
sensors, industry control and measurement SW/HW/firmware,
digital data processing, and solar energy related technologies.
He has published a number of papers and has a U.S. Patent
in industry process anomaly detection. He has been an active
contributor for 4 IEC and IEEE technical standards. He is a
member of IEEE HVTT (High Voltage Test Technology)
subcommittee and CIGRE D1.03 WG and was a member of
Technical Advisory Groups of US National
Committee of IEC.
Currently, Mr. Zhang is with Stirling Energy System Inc.
as a Principal System Engineer.

D. Bruce Osborn, President & CEO,

Stirling Energy Systems, Inc. – Mr.
Osborn has been with SES since 1999
and led one of the most successful
solar energy programs where the
product still holds the world’s record
for solar-to-electricity conversion
efficiency.
Mr. Osborn was previously the
Senior Vice President Programs &
Engineering for the Power Systems