Ancillary Services Provided from DER

with Power Electronics Interface

Xiaoyan Yu, Student Member, IEEE, Leon M. Tolbert, Senior Member, IEEE

Abstract−Distributed energy resources (DER) are quickly Power electronics offer significant potential to improve
making their way to industry primarily as backup generation. the local voltage regulation of the grid that will benefit
They are effective at starting and then producing full-load both the utility and the customer-owned DG source. Power
power within a few seconds. DER can alleviate the burden on
electronics offer the conversion of real power to match the
the utility grid by providing ancillary services, which also
helps in justifying the installation cost for the DER owner. system voltage and frequency, but this interface could do
This paper describes ten types of these ancillary services, much more. Power electronics could be designed to
which DG can provide to the utility grid. Of these 10 services, incorporate voltage and frequency conditioning for the
the feasibility, control strategy, effectiveness, and cost benefits utility. Also, various controls could be built into the power
are all analyzed for the future utility power market. electronics so that a DG system could respond to special
events or coordinate its operation with other DG sources on
Index Terms – ancillary services, distributed energy, power
the distribution system.
electronics interface.
In this paper, we investigate the possibility of using
I. INTRODUCTION DG to provide the following ten ancillary services:
• Voltage Control
D istributed generation (DG) applications currently are
primarily designated for backup and peak power
• Frequency Regulation
• Load Following
shaving conditions. Distributed generation includes
• Spinning Reserve
microturbine generators, internal combustion engines
• Supplemental Reserve (Non-spinning)
(ICEs), and fuel cells. Frequently, these generators sustain
long periods at an inoperative state until the needs of the • Backup Supply
load or the local utility require additional generation. Thus • Harmonic Compensation
DG is costly to install, maintain, and operate for most • Network Stability
commercial customers. • Seamless Transfer
DG is cost-effective only in some niche markets where • Peak Shaving
the electricity cost is extremely high, such as Hawaii and To provide the ancillary services from DG, also
the Northeast, or where outage costs are high. Two referred to as distributed energy resources (DER), the
possibilities for achieving cost-effectiveness for DG are system should contain two subsystems: one is the on-line
reducing the capital and installation costs of the systems, detection subsystem, which detects the need for ancillary
and taking advantage of additional ancillary services that services and gives a signal when an ancillary service is
DG is capable of providing. needed; the other is a function realization subsystem, which
A market for unbundled services (ancillary services) works to provide the related ancillary services after
would promote installation of DG where costs could not be receiving that signal.
justified based purely on real power generation. The For example, the same DG could be a voltage supporter
provision to produce ancillary services with DG would or a harmonics compensator. The only difference is in the
greatly alleviate the present demands on an aging power control part, so when the detection subsystem finds the
grid. Unfortunately, the demand on the grid is increasing at harmonics are beyond the preset range, then it controls the
a faster rate than improvements are being made. operation of the DG to work as a harmonics compensator;
Deteriorating power quality is a result of the imbalances in when the detection subsystem find the voltage is beyond
the grid. Ancillary services can be the bridge between the the preset range, then it controls the DG to operate as a
capabilities of DG and the needs of the utility. voltage supporter.
_____________________________
A control hierarchy must be established such that the
DG provides the most important ancillary service when
X. Yu and L. M. Tolbert are with the Department of Electrical and more than one ancillary service is needed. To decide which
Computer Engineering, The University of Tennessee, Knoxville, TN one is the most important ancillary service, we need to
37996-2100 USA (e-mail: [email protected], [email protected]). This work
was supported by Oak Ridge National Laboratory. consider economic issues, system reliability, and technical
feasibility.
 For most of the ancillary services, we only need Set the rated voltage to be 4160 V. Choose [3952, 4368]
additional software code to provide it from DG. However, to be a preset acceptable voltage range; when the voltage is
if an external energy storage capability was available, the outside this range, the voltage supporter is started. Fig. 3 is
power converter could provide high, short-duration power the comparison between the compensated voltage and the
to start motors, which is important in providing ancillary original voltage, where the voltage before compensation is
services such as load following and frequency regulation 3620 V (13% less than the rated voltage).
[1].

II. TYPES OF ANCILLARY SERVICES

Each of the ten types of ancillary services is described
in this section along with simulations to demonstrate how
these services might be provided from DG.
A. Voltage Control
Voltage Control is the injection or absorption of reactive
power by generation and transmission equipment to Fig. 2. Voltage control system scheme under grid fault condition.
maintain transmission system voltages within required
ranges or maintain the bus voltage of critical or sensitive 8000

VL with comp
loads [2]. Simulations will demonstrate how DG could 6000

perform the voltage control function. 4000

The analysis scheme for voltage control is shown in Fig. 2000

1. There are three types of conditions that may cause

VL ( V )
0
voltage sag/swell for the bus voltage of critical or sensitive
loads: -2000

• First condition: the load has some inductive -4000

component. The inductive component will not -6000

absorb active power; however, it absorbs reactive -8000

VL without comp

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
power, which will cause larger transmission time(sec)

voltage drop and a lower load voltage VL. Fig. 3. Compensation effect when VL=3620 V.
• Second condition: the real power of the load
The active and reactive current provided by DG in this
changes. Changing the impedance of the load
condition is shown in Fig. 4. Fig. 4(a) is the entire current
could simulate this condition.
provided by DG, Fig. 4(b) is the active component, and
• Third condition: a fault in the grid makes Vg
Fig. 4(c) is the reactive component.
change, so VL changes accordingly.
Generally speaking, the second and third condition may 1000
iC ( A )

cause voltage sag or swell, and the first condition only 0

causes voltage sag. -1000

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
(a) time(sec)
Vg VL 1000
RS XS
iCa ( A )

0
Utility
Line
-1000
Load 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
(b) time(sec)
1000
DG
(Voltage Support)
iCr ( A )

-1000
Fig. 1. System scheme for voltage support. 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
(c) time(sec)

Fig. 4. Current provided by DG when VL = 3620 V, (a) compensated

One of the general control methods that can realize the current, (b) compensated active current, (c) compensated reactive
basic principle of voltage support is as follows: compare current.
the actual RMS value of VL and the reference value (for
From Figs. 3 and 4, the voltage can be regulated to the
example, 480 V), and then use the difference to control the
rated voltage, and reactive current is provided by the DG
DG to provide/absorb reactive power.
only after about one cycle of dynamic process. Since the
Take a grid-fault condition as an example to show the
dynamic process is very short, DG provides almost no
simulation results. The simulation scheme is shown in Fig.
energy during the dynamic process. In fact, the dynamic
2.
process is mostly a result of the calculation period. It needs
some time (1 cycle in this simulation) to calculate the (1/60 s).
active and reactive current; and at the same time, the RMS
C. Load Following
value of the voltage VL also needs 1 cycle to be calculated
in this simulation. Fig. 6 is the system scheme for load following. DG sells
From the simulation results, we could see that, by some of power to the utility, while at the same time it
comparing the RMS value of VL to its expected value that supplies the load and tracks changes in customer needs.
controls the amount of reactive current provided, DG could Vg
RS XS ig VL
support the voltage well under the condition resulting from
a grid fault. In addition, from the simulation results of Utility
Line
inductive load and real power load changes, we can also Load

get similar results. Thus, this control method works well

under all these voltage control conditions. DG
(Load Following)

B. Frequency Regulation
Frequency Regulation is the use of online generation Fig. 6. System scheme for load following.
units that are equipped with governors and automatic For load following, part of the function is to track the
generation control and that can change in a timely fashion load, which is similar to frequency regulation. Load
to regulate frequency [2]. following is often mentioned together with frequency
When the mechanical power and electrical power of the regulation; they both address the temporal variations in
synchronous generator is not in balance, then the frequency load.
in the grid will change, and the corresponding equation is: In this system (providing ancillary services from DG),
4π H ⋅
(1) the difference between frequency regulation and load
f = P mu − P eu
ωe following is that: for load following, DG sells some of
power to the utility and tracks the load’s change at the
where Pmu is the mechanical power (W), and Peu is the same time, but for frequency regulation, DG only tracks the
electrical power (W), H is per unit inertia constant (s), ωe is load’s change. However, in some systems, the distinction
the electrical rotational speed, and f is the frequency, between load following and frequency regulation is only
ωe=2πf. Since f changes only slightly, we can treat 4πH/ωe the time periods over which these fluctuations occur.
as a constant K. In this simulation, K = 50,000. Frequency regulation responds to rapid load fluctuations
Suppose that the load is in balance with generation (on the order of one minute), and load following responds
supply before time t=0, and the frequency is 60 Hz. At time to slower changes (on the order of five to thirty minutes)
t=0, there is a step increase of 100 kW in the load power. [3]. Therefore, their precise definitions vary from system to
DG is controlled to supply this increased amount. system.
Comparing the frequency with and without regulation, the
result is shown in Fig. 5. D. Network Stability
When trying to improve transmission system use, a key
200
assumption is that the existing system stability will be
compensated power
150 maintained. Network Stability is the use of special
P (kW)

100 equipment at a power plant (power system stabilizers or

50
load increment
dynamic resistor) or on the transmission system (such as
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
DC lines, FACTs, energy storage) to help maintain
(a) time(sec) transmission system reliability [2]. For this example
(providing ancillary services from DG), we use DG to help
60
59.98 maintain transmission system reliability.
59.95
To maintain network stability, the power system needs
f (Hz)

59.9 frequency with comp

59.85
frequency without comp to have adequacy, which is defined as the ability of the
59.8 power system to meet energy demands within component
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
(b) time(sec) ratings and voltage limits. Energy storage can be used to
Fig. 5. Compensation effect for regulation, (a) load increment and enhance operation of the transmission system power flow
compensated power, (b) frequency with and without compensation. control equipment by supplementing the ability of this
Fig. 5(a) describes the comparison between the load equipment to generate or absorb active power. Therefore,
increment and the power that the DG provides. Fig. 5(b) to maintain network stability, the energy storage needs
shows the comparison of the frequency with DG must be fully available very quickly (on the order of one
compensation and without DG compensation. In this cycle).
simulation, the frequency could be controlled to be in the Network stability is similar to frequency regulation, but
preset range (59.98 – 60.02 Hz) in less than one cycle it requires a more rapid response time. DG with power
electronics could perform network stability function by F. Supplemental Reserve (Non-spinning)
monitoring frequency fluctuations and controlling the DG Supplemental Reserve (Non-spinning) is the use of
import/export, since it can respond very quickly. As shown generating equipment and interruptible load that can be
in Fig. 5, by using DG, the frequency could be controlled fully available to correct for generation/load imbalance
to be in the preset range (59.98– 60.02 Hz) in less than one caused by generation or transmission outages [2].
cycle (1/60 s), which is qualified for the network stability Supplemental reserve differs from spinning reserve
requirement. because supplemental does not need to respond to an
E. Spinning Reserve outage immediately. Traditional non-spinning reserve
needs to be available within 10 minutes.
Normally, Spinning Reserve is the use of generating
If a DG system is not on-line, it needs some time to start.
equipment that is online and synchronized to the grid such
• A microturbine requires approximately 2 minutes
that the generating equipment can begin to increase output
to start up.
delivery immediately in response to changes to
interconnection frequency, and be fully utilized within • For an internal combustion engine, the time
seconds to <10 minutes to correct for generation/load necessary to start is from 2 – 5 seconds.
imbalances caused by generation or transmission outages • For a fuel cell, most have a start-up time of
[2]. Most on-line DG could perform spinning reserve and approximately 5 minutes.
respond in less than 10 seconds. For most DER, 10 minutes is an adequate time to start,
For DG with a power electronics interface to provide and it is conceivable that an off-line DG system is
spinning reserve, the control method is to control the appropriate to perform supplemental reserve.
current from DG to let it be an active current, and by G. Peak Shaving
changing the amplitude of the current, change the DG’s Peak Shaving is the use of generation equipment during
output active power. certain peak load periods. Customers must purchase power
As an example, DG with a power electronics interface is at a higher cost during peak load periods, and the demand
simulated. Assume that at a time of 0.4 s, a frequency drop charge is a function of the peak demand at any time during
is detected, and a decision was established to use DG to a year; therefore, utilization of generation for peak shaving
supply active power. The original source provides less can reduce a customer’s operational costs. DG could
active power, and then the grid frequency will be normal perform the peak shaving function.
again. Fig. 7 is the simulation result. Fig. 7(a) is the active Peak shaving provides active power to the user during
power Psa and reactive power Psr the source provides; Fig. peak load. Because this is similar to spinning reserve in the
7(b) is the active power Pca and reactive power Pcr that the supply of active power, the accomplishment of the two
DG provides; Fig. 7(c) is the active power PLa and reactive ancillary services is similar. However, spinning reserve is a
power PLr the load absorbs. service provided to the utility, and peak shaving is a service
for the DG owner.
2000
PSa
H. Backup Supply
PS (kW)

1000
PSr
0 Backup Supply is a service that customers would
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
(a) time(sec) purchase to protect against forced outages by the
1000
generating units that provide their energy or against loss of
PC (kW)

PCa
500 transmission between their normal supply and their load
0
PCr
[2]. Unlike spinning reserve and supplemental reserve,
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
(b) time(sec) which are system services required for reliability, backup
2000
PLa
supply is a commercial service that supports individual
transactions. DG could perform backup supply.
PL (kW)

1000
PLr The basic principle is that during forced outages of the
0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 utility, DG is used to supply the load. The method is to
(c) time(sec)
control the output voltage of the DG to give the load
Fig. 7. Simulation results for spinning reserve, (a) source power, (b)
compensated power, (c) load power.
uninterruptible supply.
For example, a DG with power electronics is used in a
From Fig. 7, the active power the source must provide simulation. The results are shown in Figs. 8 through 11 for
changes from 1151 kW to 836 kW, and the DG provides four different cases. Ideally the algorithm is able to detect
this difference (315 kW). It only needs about 0.02 seconds an outage very quickly.
(from t=0.04 s to t=0.06 s) for the system to finish the For an outage that occurs at t=0.0375 s, when the
dynamic process. Obviously this short time is much smaller voltage is at its peak amplitude, the outage is detected and
than 10 s, thus on-line DG with power electronics interface then the DG begins to supply the load (Fig. 8). In fact, this
is qualified for supplying spinning reserve. is the worst condition, since the source voltage is lost when
the load voltage is at its maximum amplitude. Under this quality, voltage imbalances, and excessive zero-sequence
condition, DG needs approximately 1/12 cycle to supply currents.
the load normally. 1) Harmonic Detection:
For an outage at t=0.05 s, when the voltage is crossing Before harmonic current compensation, harmonic
through 0, it is detected and then the DG starts to supply current must be detected. There are many methods to detect
the load (Fig. 9). Under this condition, there is almost no harmonic current. We use instantaneous active current
transition process, and the DG could supply the load calculation theory to separate the harmonic component and
normally immediately. fundamental reactive component from the fundamental
For an outage at t=0.04 s, when the voltage is not 0 and active component.
not at the maximum amplitude, it is detected and the DG Instantaneous active power [4] is defined as the time rate
starts to supply the load (Fig. 10). Under this condition, of energy generation, transfer, or utilization, where M is the
there exists a transition; however, it is less than 1/12 cycle. phase number of the system:
By decreasing the value of the inductance, which M M
connects DG and the load, the transition process may be p(t ) = ∑ pi (t ) =∑ vi (t )ii (t ) (2)
short and almost disappear; however, the voltage waveform i =1 i =1
will not be as smooth as before. Take an outage that The nonactive current/power definitions are based on
happens at t=0.0375 s as an example, since this is the worst this definition and an extension of Fryze’s idea of
condition, and the result is shown in Fig. 11. nonactive current/power [5]. The definitions of
x 10
4 instantaneous active current ip(t) and nonactive current iq(t)
1
are:
0.5
i p (t ) = VPL2 ((tt)) v p (t )
VL ( V )

0 p
(3)
-0.5
iq (t ) = i (t ) − i p (t )
-1
0 0.01 0.02 0.03 0.04 0.05 0.06
time(sec)
0.07 0.08 0.09 0.1 where PL(t) is the average power of p(t) over the interval
Fig. 8. Outage occurs at ts=0.0375 s. [t-TC, t]:
t

1
x 10
4
PL (t ) = Tc1 ∫ p(τ )dτ
t −Tc
(4)
0.5

and vp(t) and Vp(t) are the instantaneous and rms values of
V (V)

the reference voltage:

L

-0.5

1 t
∫
-1
0 0.01 0.02 0.03 0.04 0.05
time(sec)
0.06 0.07 0.08 0.09 0.1
V P (t ) = v P2 (τ ) dτ (5)
Fig. 9. Outage occurs at ts=0.05 s.
TC t −TC

Using this method and selecting the reference voltage to

4

1
x 10
be the fundamental of the source voltage, the calculated
0.5 active current will be a current that is in phase with the
fundamental component of the source voltage.
VL ( V )

-0.5 2) Harmonic Compensation:

-1
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 The principle for harmonic compensation is: first,
time(sec)

Fig. 10. Outage occurs at ts=0.04 s.

calculate the fundamental current if from the load current iL,
then subtract if from the load current iL to get the harmonic
x 10
4 current, which is the current ic that the compensator should
1
provide. The system scheme is shown in Fig. 12.
0.5
VL ( V )

0 iS V iL
-0.5 Source
-1
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
time(sec)
ic
Fig. 11. Outage occurs at ts=0.0375 s with less output
inductance. load
compensator
I. Harmonic Compensation
Fig. 12. Harmonic compensation schematic diagram.
Harmonic Compensation is the use of online generation
equipment to compensate for harmonics caused by non- Fig. 13 is the compensation result. Fig. 13(a) shows
continuous loads. Harmonics can cause poor power the source current iS1 (before compensation) and the
fundamental current if. Obviously iS1 has harmonic owner.
components and does not match if. Fig. 13(b) shows the Backup supply is a service that customers would
source current iS2 (after compensation) and the fundamental purchase to protect against forced outages at the generating
current if. Fig. 13(c) shows the calculated harmonic current units that provide their energy, or against loss of
ih and the compensated harmonic current ic. We can see transmission between their normal supply and their load.
that the compensated harmonic current ic is almost exactly Thus backup supply is a commercial service that supports
equal to the calculated harmonic current ih and after individual transactions and provides the most impact to the
compensation, the source current iS2 matches the DER owner.
fundamental current if very well. Peak shaving is the use of generation equipment during
certain peak load periods. Customers must purchase power
at a higher cost during peak load periods; therefore, use of
and i ( A )

500
generation for peak shaving can reduce customers’
f

0
-500 operational costs. Thus, peak shaving provides the most
S1
i

0.35 0.4 0.45 0.5

(a) time(sec) impact to the DER owner.
Load following is the use of online generation
and i ( A )

500
equipment to track the changes in customer needs. This
f

0
-500 minimizes the power purchased from the utility. Thus load
S2
i

0.35 0.4
(b)
0.45 0.5
time(sec) following provides the most impact to the DER owner.
200 When DER transfers from stand-alone mode to grid-
i and i ( A )

connection mode, or from grid-connection to stand-alone

0
c

mode, it is expected to transfer seamlessly. By providing

h

-200
0.35 0.4 0.45 0.5
uninterrupted service to a company’s loads, seamless
(c) time(sec) transfer provides the most impact to the DER owner
Fig. 13. Simulation results for harmonic compensation, (a) source
current iS1 (before compensation) and the fundamental current if , (b) C. Provide the most impact to both the grid and the DER
source current iS2 (after compensation) and the fundamental current if owner
, (c) calculated harmonic current ih and the compensated harmonic
current ic.
Voltage control, frequency regulation, and harmonic
compensation impact the grid and the DER owner more or
J. Seamless transfer less equally.
When DG transfers from stand-alone mode to grid- Voltage control regulates a bus voltage, which
connection mode, or transfers from grid-connection mode improves the power quality of the DER owner, but it also
to stand-alone mode, it is expected to transfer almost minimizes reactive power drawn from the grid, improving
instantaneously, which is called seamless transfer. A good system efficiency and stability.
reference that contains algorithms for seamless transfer can For frequency regulation, the goal is to maintain the
be found in [6]. In a wider definition, seamless transfer is rated frequency of the grid by tracking the load’s demand.
the ability for online generation to transition among various The DER owner is also a part of the load, so frequency
ancillary services without power delivery disruption. regulation provides impact to both the grid and the DER
owner.
III. BENEFITS TO GRID AND CUSTOMER Through harmonic compensation, the source avoids the
By the goal of the ancillary services, these ten types of need to provide harmonic current to a load with harmonic
ancillary services could be divided into those that mostly components, because it has been provided by the DER.
benefit the grid, those that mostly benefit the DER owner, Thus there will be no harmonic current flowing through the
and those that benefit both more or less equally. rest of the grid. This improves system efficiency and
stability by reducing losses in the rest of the system. It may
A. Providing the most impact to the grid also enable the owner of a harmonic-producing load to
Spinning reserve, supplemental reserve (non-spinning), avoid paying penalties to the utility for harmonic-
and network stability provide the most impact to the grid. producing loads that do not comply with IEEE 519,
All three of these ancillary services aim to keep the load “Harmonics Control in Power Systems.”
and supply in balance, which keeps the frequency of the
utility at its normal value (60 Hz), so they are system IV. ADDITIONAL CONSIDERATIONS WHEN USING DG FOR
services required for reliability and provide the most ANCILLARY SERVICES
impact to the grid. Reliability and quality are the two most important facets
of any power delivery system. In recent times, the issues
B. Providing the most impact to the DER owner involved with power quality issues and custom power
Backup supply, peak shaving, load following, and solutions have generated a tremendous amount of interest
seamless transfer provide the most impact to the DER among power system engineers. Since these power quality
problems bring so much loss to utility users, ancillary B. Interconnecting DGs together
services must be provided to solve them.
This is another way instead of using external energy
Providing ancillary services from DER will be a good
storage devices to produce more power to supply. DG with
solution to power quality problems, which could be seen
power electronics is not difficult to connect with each
from the following points:
other, since most have a DC link. So by using the common
• Local regulation is much more efficient with local DC link, DGs will act just like one DG with a higher power
sources, and the DER can supply precisely the level of rating [9].
regulation needed [7]. Here the regulation not only
means the regulation of voltage, but also the regulation C. Use of advanced semiconductors
of frequency (network stability, load following, and
frequency regulation). Therefore, DER is quite suitable Some cost reductions may also be realized from
to provide network stability, load following, and advances in semiconductors. IGBTs are currently leading
frequency regulation. the market for semiconductors used in medium-level (up to
750 kVA) Power Conversion Systems. IGBT technology
• For harmonic compensation and network stability, has wide industry support and is most likely to provide the
both of them require fast response ability; the DER greatest cost reduction potential for the near future [8].
with power electronics is quite appropriate to provide Recently, silicon carbide- (SiC-) based power devices
these two ancillary services. have been drawing increased attention because of their
• For backup supply and peak shaving, DER is perfect superior characteristics compared with silicon- (Si-) based
to provide them because of its proximity to the user. In power devices. A SiC-based power converter would have
fact, distributed generation applications today are the benefits of reduced losses, higher efficiency, up to 2/3
primarily designated for backup and peak power reduction in the heatsink size, smaller passive components,
shaving conditions. and less susceptibility to extreme ambient heat. DG power
• DER could be reserve to the utility, and perform other converters would certainly benefit from utilizing SiC-
ancillary services when the utility has enough real power devices.
power. Thus, there will be no waste for the resource.
V. CONCLUSIONS
There are several recommendations in support of
DER is gaining ground in the area of power quality
providing ancillary services from DER:
improvement. Increasingly, digital loads that demand high
A. Ability to interface with energy storage devices power quality are showing up on the electrical network. A
momentary outage of a few cycles that used to be of no
The ability of accepting power from a storage device concern can now be disastrous to digital loads such as
such as a battery or ultra capacitor will help DG to provide semiconductor plants. The outage costs alone can range
a large amount of active power. For some of the ancillary from thousands to millions of dollars; thus outage
services, such as Voltage Sag Support and Harmonic avoidance can make it cost-effective to install onsite
Compensation, only reactive power is supplied, so they do generation at the load.
not need an external energy storage device. However, if Unfortunately, most DER systems are not designed for
external energy storage was available, the DG could uninterruptible power supply operation. The integration of
provide very high, short-duration power to start motors, power electronics, DER, and energy storage could
which is very important in providing ancillary services ultimately be the answer for power quality. However, this
such as load following and frequency regulation. integration will not occur until the cost of DER is lowered,
Battery energy storage (BES) and superconducting along with the cost of its power electronics.
magnetic energy storage (SMES) are the most universally A market for unbundled services (ancillary services)
applicable forms of energy storage for transmission would promote installation of DG where costs could not be
applications. Both have relatively high efficiency, few justified based purely on real power generation.
requirements for sitting, and are capable of storing energy Power electronics offer the conversion of real power to
in amounts appropriate for transmission applications. There match the system voltage and frequency, but this interface
are many other forms of energy storage, including could do much more. This paper describes ten types of the
capacitive energy storage (CES), flywheel energy storage ancillary services, which DG with power electronics
(FES), compressed air energy storage (CAES), pumped interface can provide. Of these 10 services, the feasibility,
hydro, etc., but these either have stringent site requirements control strategy, effectiveness, and cost benefits are all
(CAES and pumped hydro), are still under development for analyzed for the future utility power market.
sizes suitable for transmission applications (FES), or For most of the ancillary services, we only need
appear to be prohibitively expensive for the energy additional software code to provide it from DG. However,
requirements of transmission systems (CES) [8]. external energy storage devices may enhance the DG
capability to provide these ancillary services.
 VI. ACKNOWLEDGMENT BIOGRAPHIES
The authors gratefully acknowledge the contributions of
Xiaoyan Yu (S 2004) received the B.E. and
Jeremy Campbell, Tom Rizy, Yan Xu, and Saijun Zhang M.S., in electrical engineering from the
for their work on the original version of this document. Huazhong University of Science and
Technology, China, in 2000 and 2003,
respectively.
V. REFERENCES She presently is a Ph.D. student in electrical
[1] J. B. Campbell, T. J. King, B. Ozpineci, L. M. Tolbert, X. Yu, Y. Xu, engineering at The University of Tennessee. Her
“Ancillary Services Provided from DER,” Oak Ridge National research interests are power quality, distributed
Laboratory Report, ORNL/TM-2005/263. energy, and the power electronics interface with utilities.
[2] E. Hirst, B. Kirby, “Creating Competitive Markets for Ancillary
Services” ORNL/CON-448, Oak Ridge National Laboratory, Oak
Ridge, TN, October 1997. Leon M. Tolbert (S 1989 – M 1991 – SM
[3] E. Hirst, B. Kirby, “Separating and Measuring the Regulation and 1998) received the B.E.E., M.S., and Ph.D. in
Load-Following Ancillary Services,” Utilities Policy, vol. 8, pp. 75- Electrical Engineering from the Georgia
81, June 1999. Institute of Technology, Atlanta, Georgia.
[4] Y. Xu, J. N. Chiasson, L. M. Tolbert, “Nonactive Current Definition Since 1991, he worked on several electrical
and Compensation Using a Shunt Active Filter,” IEEE International distribution projects at the three U.S.
Conf. on Harmonics and Quality of Power, Lake Placid, New York, Department of Energy plants in Oak Ridge,
Sept. 2004, pp. 573 – 578. Tennessee. In 1999, he joined the Department
[5] H. Akagi, Y. Kanazawa, A. Nabae, “Instantaneous Reactive Power of Electrical and Computer Engineering at the
Compensators Comprising Switching Devices without Energy University of Tennessee, Knoxville, where he is
Storage Components,” IEEE Trans. Ind. Appl., vol. 20, May/June presently an associate professor. He is an adjunct participant at the Oak
1984, pp. 625-630. Ridge National Laboratory. He does research in the areas of electric power
[6] R.Tirumala, N. Mohan, C. Henze, “Seamless Transfer of Grid- conversion for distributed energy sources, reactive power compensation,
Connected PWM Inverters between Utility-Interactive and Stand- multilevel converters, hybrid electric vehicles, and application of SiC
Alone Modes,” IEEE Applied Power Electronics Conference and power electronics.
Exposition (APEC), 2002, pp 1081 – 1086. Dr. Tolbert is a registered Professional Engineer in the state of
[7] J. D. Kueck, B. J. Kirby, L. M. Tolbert, D. T. Rizy, “Tapping Tennessee. He is the chair of the Educational Activities Committee of the
Distributed Energy Resources: Reactive Power Is the Key to an IEEE Power Electronics Society and chair of the Special Activities,
Efficient and Reliable Grid,” Public Utilities Fortnightly, vol. 142, Industrial Power Conversion Committee, IEEE Industry Applications
no. 9, September 2004, pp. 47-51. Society. He is the recipient of a National Science Foundation CAREER
[8] S. Eckroad, “FACTS with Energy Storage: Conceptual Design Award and the 2001 IEEE Industry Applications Society Outstanding
Study”, EPRI, CA, TR-111093, December 1999. Young Member Award.
[9] B. Ozpineci, L. M. Tolbert, Z. Du, “Multiple Input Converters for
Fuel Cells,” in Proc. 2004 IEEE Industry Applications Society
Annual Meeting, pp. 791-797.