IEEE TRANSACTIONS ON SMART GRID, VOL. 1, NO.

3, DECEMBER 2010 225

The Next Generation of Power Distribution Systems

G. T. Heydt, Life Fellow, IEEE

Abstract—This paper summarizes diverse concepts for the next

generation of power distribution system. The objective is to bring
distribution engineering more closely aligned to smart grid philos-
ophy. Issues of design, operation, and control are discussed with
regard to new system theoretic as well as component/materials ad-
vances. In particular, two transmission engineering techniques are
modified for use in distribution engineering: state estimation, and
locational marginal pricing. The impact of electronic control in
distribution systems is discussed. Because education and training
have a great impact on distribution engineering, these topics are
discussed as well.
Index Terms—Direct digital control, distribution engineering,
energy management, locational marginal prices, power delivery,
power distribution systems, power engineering training, smart
grid, state estimation.

I. POWER DISTRIBUTION ENGINEERING

F OCUS ON THE smart grid has brought renewed interest

in enhancement of all parts of power systems. The “dis-
tribution system” refers to the lower voltage (e.g., 40 kV) Fig. 1. Radial, looped, and networked primary and secondary systems.
system that is energized at the distribution substation, and de-
livers power to loads via a primary distribution system. Tradi-
tionally, the primary distribution system has been a radial net- there would be a transition from mechanical devices and tra-
work with many feeders and laterals energized by these feeders. ditional materials used in components to advanced materials;
However, there are distribution systems that are networked. The and there would be application of system theory to obtain the
literature of distribution engineering is very large, and [1]–[5] best performance from the given assets [7]. As examples, man-
will serve as samples of the classical material. agement/control of power is proposed versus the uncontrolled
The essence of the smart grid philosophy is to implement the action of the Kirchhoff laws; utilization of sensory information
seven principles [6]: on load demands and availability of renewable generation
• self-healing from power disturbance events; resources is proposed; and automation of switching in the
• enabling active participation by consumers in demand re- distribution system is envisioned (versus the legacy procedure
sponse; of operator switching). These are the focus of this paper.
• operating resiliently against physical and cyber attack;
II. DISTRIBUTION SYSTEM CONFIGURATION
• providing power quality for 21st century needs;
• accommodating all generation and storage options; The design configuration of the distribution system has direct
• enabling new products, services, and markets; bearing on its operation and performance. Traditionally, a higher
• optimizing assets and operating efficiently. voltage subtransmission or transmission system energizes sub-
In many ways, these points are appropriate to distribution station transformers whose secondaries energize “primary dis-
engineering design and operation. The generic topics listed tribution feeders” in a radial configuration (see Fig. 1). In this
effectively mean that sensory information would be used for configuration, control of power flow is minimal, but the cir-
automatic control and operation of the distribution system; cuit design is straightforward. Laterals are used to take power
from main feeders. Protection in the system is mainly relegated
Manuscript received August 04, 2010; accepted September 14, 2010. Date of
to fuses that are coordinated to give protection of equipment
publication November 01, 2010; date of current version November 19, 2010. with minimum service outage. Service to the load is provided
This work was supported in part by the Power Systems Engineering Research through distribution transformers that are usually protected by
Center (PSerc) under National Science Foundation Grant EEC-0001880 and
Grant EEC-0968993 and in part by the Future Renewable Electric Energy Distri-
a fuse. In the United States, on the average, about 5 to 10 cus-
bution and Management Center (FREEDM) under NSF Grant EEC-08212121. tomers are served at the residential level by a single distribution
Paper no. TSG-00103-2010. transformer.
The author is with the Department of Electrical, Computer, and Energy Additional performance can be obtained from the basic ra-
Engineering at Arizona State University, Tempe, AZ 85287 USA (e-mail:
Heydt@asu.edu). dial configuration with a double feed from two diverse sources,
Digital Object Identifier 10.1109/TSG.2010.2080328 switched to obtain power from a sound source. The IEEE Red
1949-3053/$26.00 © 2010 IEEE
226 IEEE TRANSACTIONS ON SMART GRID, VOL. 1, NO. 3, DECEMBER 2010

and Gold Books [8], [9] describe the load types for which a
double feed (excess service capacity) is required. Mechanical
switches can also be used for service restoration in radially con-
figured systems [10], [11] but electronic switches are readily
designed for 15 kV service in which only closing (not opening)
is required, and these electronic switches can operate in the sub-
cycle range.
When load density is high, and/or when higher reliability
is desired, the use of networking is used in distribution sys-
tems. Networking the primary feeders is not unusual in city
centers. Networking may require more elaborate protection and
switching provisions in order to obtain the best performance.
Fig. 1 shows a networked primary system. An alternative is to
network the secondary system (i.e., at the lower voltage distribu-
tion transformer secondary). Networking the secondary system
results in high reliability, but there are disadvantages of required
excess transformer capacity and more complex protection re-
quirements. Reference [12] contains a comparison of networked
distribution systems.
It is envisioned that many future distribution systems will be
networked or looped. It is also envisioned that sensors will be Fig. 2. A comparison of energy losses in transmission and distribution systems,
used for direct control of system elements, more than in con- with data taken from various sources, taken at 75% loading operation.
temporary distribution systems. The main points in future dis-
tribution system designs are: autoconfiguration capability (e.g., designs, the capacitor current can be pulse width modulated to
electronic switch closures to reconfigure systems such as the use obtain smooth control and low harmonic distortion. The latter
of electric ship technologies in reconfiguration [13], [14]); more is important in compliance with IEEE Standard C18 [15]. The
flexible and managed integration of renewable resources in the applications that usually favor subcycle control include: ac
distribution system; implementation of reliability management electric arc furnaces, large highly varying industrial loads such
(e.g., increased reliability where needed, cost/benefit analysis to as steel rolling mills [16], certain mining loads (e.g., “plasma
identify priority circuits, possible implementation of hyper-reli- torches” and drag lines). The traditional approach with regard
ability philosophies where warranted); management of reactive to objective of reactive dispatch has been to render the power
power; improved customer average service reliability; accom- factor at the high side of the distribution substation transformer
modation of electric vehicle loads; development of strategies for to be a few percent leading—this giving reactive power support
when to network legacy radial systems. to the transmission system.
Energy losses in the distribution system primaries are de-
III. DISTRIBUTION SYSTEM PERFORMANCE pendent on circuit loading and circuit design, but are generally
below 7% of the energy delivered (see Fig. 2). Note that the chal-
lenge of electronic control is for designs in which the losses in
A. Nominal Operation
the electronic switches are lower than those being saved by im-
Operation and evaluation of distribution systems may be plementing the switching. The trade-offs in losses may be offset
divided into voltage regulation, efficiency, reliability/avail- by a variety of benefits of electronic switching: for example,
ability, power quality, operation under faulted conditions, electronic control to reduce flicker (note that the threshold of
and high-voltage issues. Relating to voltage regulation, the human perception of flicker from an incandescent lamp is in the
traditional approach has been to render voltage drops in the order of that produced by %). The subject of dis-
system to be low by reactive power management. This is tribution transformer losses is discussed in Section IV.
often accomplished with fixed shunt capacitors and capacitors Losses and efficiency, as indicated above, are nominally cal-
with some kind of control (e.g., radio dispatched, voltage culated in the distribution system in the 7% range or less. If sec-
dispatched, timer control). The next generation of distribution ondary losses are included, perhaps an additional 1% loss occurs
system permits the use of electronic controls for reactive power (loosely speaking, both energy and power, and medium load).
dispatch: this can be accomplished by electronically switching In many countries, the nominal secondary distribution voltage is
capacitors (e.g., a distribution class static var compensator); 220 V, and this offers about a 4:1 reduction in lower secondary
controlling reactive power demand at loads; and through the conductor loss versus 120 V designs. In 120 V systems, it ap-
utilization of in-line controllers. In all cases, the cost of the pears that some residential and commercial loads are migrating
electronic control needs to be weighed versus the benefit. to 220 V ratings—and this is favored from a loss point of view.
For example, is high-speed electronic control (e.g., subcycle) The safety issues might be resolved via the use of ground fault
warranted? In the 15 kV distribution class, shunt capacitors interrupters of modern design. In the distribution primaries, in
are commonly in the 0.5–5 MVAr range. This application many venues in North America, legacy 5 kV class systems are
generally results in switched currents below 45 A. In electronic largely converted to 15 kV class—and there is a clear trend to
HEYDT: THE NEXT GENERATION OF POWER DISTRIBUTION SYSTEMS 227

Fig. 3. Primary distribution system networking.

35 kV designs [22]. It appears that for moderate and high density TABLE I
circuit loading (e.g., greater than about 80 MVA/km ), 35 kV EQUIVALENT TIMES OF FAILURE AND REPAIR OF SERIES AND PARALLEL
COMPONENTS
circuits may be favored—especially when the number of points
of delivery are limited. Such a conversion will give more than
5:1 loss reduction in the primary conductors over 15 kV designs.
From the point of view of electronic controllers, the 35 kV class
is a challenge: the first challenge is the stacking of electronic
controls to obtain the nominal 29 kV peak inverse voltage (three
phase application, safety factor not included), and the second
challenge is to obtain an acceptable basic impulse level in the
semiconductor switches (e.g., 200 kV for some 35 kV circuits). of hyper reliability, it may be expedient to express reliability in
Reliability and Availability Issues: Reliability and avail- terms of “the number of nines” defined as
ability of electric power from distribution circuits is a key
part of distribution engineering. The conventional measures of (1)
reliability have been the System Average Interruption Duration
where and refer to the probability of normal operation and
Index (SAIDI) and the System Average Interruption Frequency
outage respectively. Thus, if . In [24],
Index (SAIFI)
duplicate feeder service is considered to raise and it is found
Total Duration of all interruptions that can be raised by two to three through the addition of an
SAIDI added distribution feeder that is independently derived.
Total number of customers connected
Total number of interruptions The following are viewed as being realistic possibilities for
SAIFI increasing the reliability of distribution systems:
Total number of customers connected
Conversion of overhead systems to underground: This con-
Other similar “count indices” have been proposed [23] and used version generally results in longer repair times in cases of
to quantify distribution system reliability. A shortfall of these failures, but longer mean times to failure. Note that if the mean
indices is that local reliability is not well quantified: that is, repair time of circuits is and the mean time to failure is
if a customer adds a distributed generator for local load sup- , and if , the series and parallel combination of
port, this often has minimum impact on the SAIDI and SAIFI circuit components results in an equivalent overall and
of the feeder since these indices are averaged over all connected as shown in Table I. From this elementary view, for inclusion
customers. It is possible to utilize an alternative index—for ex- of underground circuits, one expects improved mean time to
ample, a Customer Average Interruption Duration Index—that failure, but longer outage duration.
focuses on customer reliability. The need for reliability indices Networking distribution primaries: Perhaps the most evident
often relates to the allocation of resources: it makes sense to al- method of dramatically improving reliability is through net-
locate resources where they are needed to bring errant feeder working (see Fig. 3). It appears that as automated switching
reliability in line with system wide reliability. But the customer and automated restoration are realized, the trend shall be to net-
also needs to know what level of distributed generation or ex- work more distribution primaries. Added intelligence (i.e., the
cess capacity of service is needed to obtain satisfactory perfor- utilization of sensory information) is used to improve service
mance for a given application. The subject of “hyper reliability” restoration by switching to minimize unserved energy after
arises in cases in which load interruption is especially critical. an outage [10], [25]. One technique for high-speed restora-
The IEEE standards address this issue in [9]. Note that in cases tion entails a table lookup algorithm in which measurements
228 IEEE TRANSACTIONS ON SMART GRID, VOL. 1, NO. 3, DECEMBER 2010

determine which row in a precalculated table applies to the tion (CBEMA) and Information Technology Industry Council
given operating condition. If an outage is observed, the table (ITIC) power acceptability curves [27]. At industrial power
gives the required switching to restore service. In the case of levels, distribution class dynamic voltage restorers (DVRs) are
networked primary systems, the reliability enhancement is so useful to obviate momentary events.
great (e.g., an order of magnitude reduction in annual average Managing reliability: It is possible to apply a system-theoretic
outage time) that there is a much lower motivation to include approach to managing reliability, losses, and cost. This is
any further forms of reliability improvement (e.g., distributed accomplished through multiobjective optimization. For ex-
generation). The obvious disadvantage of networking is the ample, an aggregate objective function is formulated as a
added complexity of protection. The networked distribution function of the control variables (assembled in the vector )
system begins to evolve to a system that closely resembles the
transmission system. The use of electronic protection devices (2)
(e.g., solid state fuses, electronic fault interruption devices)
and improved fault detection methods and sensors will make where weights are used to render the component functions in
coordination more effective, but protection is nonetheless a common units and to reflect a management plan. In (2),
hobble to networking. is used to capture the reliability of a given plan is
Networking distribution secondaries: Some city centers employ used to capture a “cost” of losses, and is used to model
networked distribution secondaries [26]. In many applications, asset costs. Then the are varied in order to obtain a collec-
the SAIDI and SAIFI improve by an order of magnitude under tion of optimal values of , denominated , for each selection
such networking. As in the case of networked primaries, pro- of . These are denominated as Pareto optimal values. If
tection must be carefully designed and coordinated. Increasing the Pareto optimal values are envisioned in a multidimensional
primary circuit voltage: In general, increasing the distribution space that spans , and the values of for which
primary voltage results is better average reliability over all variation in the gives minimum displacement of the Pareto
customers, mainly due to the added MWh energy delivery optimum are selected, a multiobjective solution is obtained. The
levels of the primary circuits. The newest cable technologies Pareto approach has been applied in power engineering, and a
readily allow 35 kV class circuits in voltage graded cables, and full description appears in [28], [29].
in many applications, overhead circuits may be upgraded.
Including generation in the distribution system: The classic B. Power Quality
solution to customer reliability problems is to employ local The power acceptability curves (e.g., CBEMA and ITIC) are
generation. The recent interest in increasing solar photovoltaic curves that partition a two dimensional plane that represents
and wind generation resources (mainly to reduce CO emis- operating conditions. The two dimensions are the depth of a
sions, to reduce energy demand from the grid, and to appeal to low-voltage (or high-voltage) event, and, secondly, the dura-
public interests) may be a way to enhance customer reliability. tion of an event. Thus, an event of short duration and small
There are cost effectiveness questions to this approach, and depression in voltage magnitude is deemed to be “accept-
the uncertainty in the availability of solar energy (“insolation”) able”; and, an event that is long in duration, and severe in its
and wind energy generally make distributed generation mainly low voltage is assessed to be “unacceptable.” This type
of value in reducing the integrated energy demand in the dis- of demarcation may be used by equipment manufacturers (es-
tribution system. The distribution system itself would usually pecially digital technologies) to determine the duration of “ride
need to be designed to accommodate the case of no wind/solar through” capability that must be accommodated. The shape of
generation. Since wind and solar energy systems are generally the demarcator in the power acceptability curves may be deter-
asynchronous, electronic converters are required to interface mined by experiment (e.g., apply various low-voltage events of
with the ac system, and these converters may be operated to different durations to equipment, and assess when percent of
assist in var management, phase balancing, and power factor the equipment operationally fails); or a model of the load may
control as a side benefit. be used to apply a system-theoretic approach to the design of the
Utilization of ungrounded and resistance grounded systems: partition. In the latter case, a standard IEEE rectifier load may be
Ungrounded and resistance grounded systems are usually per- assumed with dc-side filter capacitors to block 120 Hz from the
mitted by codes provided the safety requirements are satisfied. dc circuit. Such a circuit is then analyzed to determine when the
Ungrounded systems have advantages in reliability, but disad- dc voltage goes below the 87% IEEE Red Book limit. This has
vantages of surge protection. In the presence of high-speed, been done in [30] to obtain a “formula” for the CBEMA curve.
subcycle electronic controls, there may be applications in which If the ultimate load is rotating, the drives used may be modeled
ungrounded and resistance grounded systems would be valued and a criterion other than per unit may be used (e.g.,
for their reliability, yet comply with protective interruption for a low machine speed limit might be used). Electronic controls
surges. now have the capability of moving the majority of low-voltage
Power conditioning: Power conditioning, including the use events to within the “acceptable” region of both the CBEMA
of uninterruptable power supplies, and devices with energy and ITIC curves.
storage capability can dramatically increase reliability. The One disadvantage of electronic switching is the generation of
electronic controls in these devices can render a momentary higher frequency currents (and subsequent voltages) in distri-
event so short that the event lies within the “acceptable” region bution secondaries and possibly in the primaries as well. The
of the Computer Business Equipment Manufacturers Associa- advent and promotion of pulse width modulation, (PWM) es-
HEYDT: THE NEXT GENERATION OF POWER DISTRIBUTION SYSTEMS 229

pecially in the carrier frequency range of kHz makes

this power quality issue minimal. Most distribution transformers
and substation transformers have low bandwidth, and high fre-
quencies (e.g., multiples of the PWM carrier frequency plus or
minus the modulating frequency) are attenuated. Additionally,
asynchronous operation of the PWM converters does not result
in the simple addition of high frequency components of current
in the primary distribution system. Some high-power PWM con-
verters also use advanced switching techniques to minimize the
high frequency impact of these loads. The use of electronic con-
verter technologies may also be used to ameliorate other power
quality problems such as three phase unbalance since individual
phase control is feasible.

C. Wide Band Gap Semiconductor Switches

The use of semiconductor switches in power distribution for
control and protection has limitations in voltage and current as
well as cost. Presently, most high-power switching applications
use insulated gate bipolar transistors (IGBTs) as the principal
switching agent—and these devices have been engineered to at Fig. 4. MEMS used as a circuit breaker. The P and N control signals elec-
least the 3 kV, 1 kA level. Gate turnoff transistors (GTOs) are tromagnetically operate all MEMS switches in the negative and positive poles
capable of doubling both the voltage and current level. At higher respectively.
power levels, thyristors are generally used. All of these minority
carrier devices can be connected in series and parallel in order sion engineering, become more feasible. The electronic solu-
to obtain much higher voltage ratings. The next generation of tions, however, may offer fast interruption as an added feature
switches are likely to come from wide band gap semiconduc- of energy management and control. Note that these devices must
tors (e.g., SiC) which could reach 20 kV in bipolar devices. As comply with basic impulse level voltage ratings. Losses in these
these devices are commercialized, it is likely that many of the electronic devices need to be considered and weighed versus
problems of mechanical switches and high-speed control would their fast interruption capabilities. Mechanical circuit breakers
be alleviated. Electric vehicle deployment has been cited as a have been commercialized down to interruption time in the tens
driver for the development of wide band gap semiconductor de- of milliseconds, and it may be possible to utilize an electronic
vices, including diamond which has a thirty times higher electric interrupter for initial interruption, followed by a mechanical
breakdown field as compared to silicon. breaker for subsequent interruption. The hybrid configuration
gives the advantage of ohmic disconnection—a clear advantage
D. Faulted Systems especially in primary circuits.
Distribution system faults are traditionally cleared by in- Although circuit interruption is an old technology, new types
suring sufficient fault current amplitude to operate a fuse. The of switches may offer revolutionary capabilities, especially in
fuses are time and current coordinated to minimize the services niche applications. An example is the use of microelectrome-
interrupted. Typical fault currents depend on the fault location, chanical switches (MEMS) to interrupt distribution current [33].
but can exceed double the rated current in secondary circuits, MEMS have excellent capability of high-speed operation, but
and well above this level in the primaries. At the distribution their conductance in the OPEN state is nonzero, and their resis-
substation, short circuit ratios (i.e., ratio of short circuit current tance in the CLOSED state is nonzero. Their voltage withstand
amplitude to circuit current averaged in 15 min intervals over capability is limited as well. Most MEMS cannot be operated
the previous year) can exceed 10. At high levels of fault current, to interrupt current. Fig. 4 shows one approach to accommo-
fuses and overcurrent relays are quite effective. However, there date these shortcomings using series and parallel combinations
is motivation to: in conjunction with both voltage and current grading resistors.
i) reduce fault current amplitude to protect solid state com- Series/parallel snubbers may be used to avoid both damaging
ponents in the distribution system; currents/voltages (not shown in Fig. 4).
ii) speed up both detection and circuit interruption to make In Fig. 4, diodes are used to separate the positive and negative
the fault event less disruptive (e.g., so that the event still current half-cycles of , and the MEMS in the positive current
lie in the “acceptable” region in a given power accept- pole are opened only during the negative half-cycle of , and
ability curve). the negative current pole MEMS are opened only during the
Both of these objectives can be accommodated using electronic positive half-cycle of . The and control signals are used
technologies. In some cases, the expense of electronic solutions to electromagnetically open the MEMS in the appropriate part
may not warrant their use, and passive solutions such as fault of the cycle of .
current limiters [31] or resonant devices may be consid- In order to obtain reliability of such a “MEMS circuit
ered. As distribution circuits migrate to networked configura- breaker,” series and parallel path redundancy are used. The
tions, relaying and the use of circuit breakers, as in transmis- number of redundant elements is calculable using Markov
230 IEEE TRANSACTIONS ON SMART GRID, VOL. 1, NO. 3, DECEMBER 2010

analysis in which different failure states for the MEMS are IV. COMPONENTS, TOOLS, ANALYSIS, AND TRAINING
modeled.
Fault interruption considerations frequently imply the impor- A. Components
tance of the speed of the interruption itself. However, as the fault There is no doubt that hardware drives physical systems, and
interruption times drop below a few cycles, the fault detection the capabilities of hardware form the basis of capabilities and
problem assumes considerable importance. Approaches to fault limitations of system performance. In the case of distribution en-
detection enhancement include the following. gineering, the reliance on mechanical switches and transformers
• The use of ground fault interrupter technology. as the main underlying system components dictates the perfor-
• The use of information to detect fault currents near mance of legacy systems. The cited components do have their
zeroes of . advantage in the sense that they have evolved to be resilient
• Process with a band limited digital filter with unit gain (e.g., to overvoltages, heat, overcurrents, mechanical damage)
and radians phase shift. and relatively low loss. However, many legacy components are
• The use of innovative signals, such as estimated voltages not well suited for high-speed operation, direct digital control,
and currents using distribution class state estimation in and placement in locations of tight physical dimensions. Im-
near real time. pressive are advances that have been made in the development
• The detection of anomalous frequency spectral compo- of materials-based devices such as metal oxide varistors ca-
nents in line currents, and the detection of chaotic com- pable of switching 70 kA in the 1.0 ns range; transient voltage
ponents to identify arcing faults [34]. surge suppressors (mainly used in electronic circuit protection);
• Detection of loss of voltage/low voltage at secondary load and “solid state fuses” which are series connected solid state
buses. switches that are operated (e.g., opened) by a fault current de-
tection circuit.
E. Basic Impulse Level
One of the most successful developments of improvements in
The basic impulse level (BIL) in a distribution system is a legacy distribution components has been in the design and com-
specified insulation level that is applied for lightning protection, mercialization of low loss distribution transformers. Reference
safety, component protection, and operational integrity. Gener- [19] summarizes the mandated efficiency standards of distribu-
ally, the BIL is coordinated in the sense that the higher BIL is ap- tion transformers manufactured in the United States. These effi-
plied at circuit components that are closest to the overhead trans- ciency mandates range from 98.62% to 99.49% for single phase
mission system (e.g., at the distribution substation), and graded liquid immersed units [37]. The losses in magnetic transformers
lower in voltage as one progresses down to the secondary cir- have been reduced to a low level which, especially when com-
cuits. The BIL specifications have evolved over the last century bined with excellent high-voltage and transient resilience, are
in response to damaging and unsafe events and to transmission difficult to replicate with alternative designs. However, the con-
and distribution engineering advances. The BIL can be as much trol possibilities of magnetic transformers are not favorable to
as 200 kV in 35 kV distribution systems [35]. Methods of re- applications in which energy management at the transformer is
ducing the impact of lightning and switching impulses/surges desired.
include: lightning arresters; metal oxide varistors; gaps of var- Additional distribution component development is exempli-
ious designs; discharge lamps; and capacitors. The designs of fied as follows.
these measures have made considerable advances and now span • Solid state transformer: essentially a dc/ac/ac controlled
a few hundred kJ in energy dissipation capability to well over converter that is capable of high-speed control of loads
200 MJ [36]. Two step and multistep arresters can provide pro- and customer connected sources (e.g., wind and solar re-
tection for a range of high-voltage conditions of various tem- sources) [38].
poral characteristics. • Solid state fault interrupter: a solid state switch in series
One unfortunate issue relating to BIL is that semiconductor with circuits to be protected.
switches often do not have substantial withstand voltage above • Distribution class unified power flow controller: a flexible
their nominal switch ratings. In the next generation of electronic ac transmission system device that is redesigned for distri-
controls, measures need to be taken to accommodate or manage bution service, capable of active and reactive power flow
the BIL requirements, and these include: controls in several different operational modes.
• development of much higher peak inverse voltage ratings • The development and use of fault current limiters [32].
for semiconductor switches;
• utilization of series connected switches to reach the re-
B. Electric Vehicle Loads
quired BIL in controlled circuits;
• utilization of innovative ungrounded connections; The prospect of electric vehicles has been on the power
• combining electronic switches with mechanical circuit engineering horizon for almost 100 years. It seems that this
breakers in a hybrid configuration; load is likely to materialize at some level including “vehicle
• redesigning the BIL coordination requirements, especially to grid” configurations which have some intelligent charging
in underground distribution; modes. The electric vehicle also offers the possibility of energy
• recasting standards and practices to allow high-speed ar- storage. The impact of electric vehicle charging on the distri-
resters of high energy capacity and multistep arresters in bution system has been variously estimated depending on the
some test conditions. type of chargers and the density of charger loads; however, the
HEYDT: THE NEXT GENERATION OF POWER DISTRIBUTION SYSTEMS 231

low end of estimates are somewhat below 5 kW per charger

during the charge cycle to over 20 kW for a fast charge. Even at
10% load penetration, the charger load would require changes
in distribution system diversity calculations and service kVA
sizing. References [56] and [57] are a small sampling of the
literature of this area.

C. Tools and Analysis Methods

Classical distribution engineering analysis tools include:
• short circuit studies;
• soad flow studies; Fig. 5. Direct control of distribution system components.
• voltage regulation studies;
• protection design, coordination, specifications, and set-
tings; combination of solid state switched device with a varistor
• mapping and visualization software; to limit fault current, or a passive series resonant
• load diversity calculations. device that employs a saturable inductor.
These tools are rapidly being augmented by software and mid- • DC subsystems in commercial buildings; and/or combined
dleware with the following capabilities: ac and dc systems with single specially designed recepta-
• performance of solar photovoltaic systems located in the cles to provide both voltages selectively.
distribution system, including transient response; The basic concept of direct control is shown in Fig. 5.
• performance of wind energy systems; Direct controls are driven by sensory measurements. Con-
• energy management tools; temporary technology allows one to make measurements that
• automated meter reading systems; are synchronized (time stamped) much like the synchrophasor
• automatically dispatched reactive power assets. [phasor measurement unit (PMU)] technology in use in trans-
In addition to the performance-based devices listed, there is a mission systems [43], [44]. In transmission systems, enhanced
focus on the development and use of sensory systems in dis- sensory signals are obtained using state estimation [45]. The
tribution engineering. As an example, the role of synchronized basis of this estimation is the use of measurements located in
measurements [39] has reached an economic point where a vector , potentially contaminated with noise or errors . The
large number of measurements might be used to drive controls measurements are related to a set of system “states” denomi-
automatically. In distribution engineering, controls refer to nated as , and the nonlinear relation
energy management (e.g., control of user permitted loads such
as air conditioners, swimming pool pumps, electric water (3)
heaters); remote dispatcher controlled energy resources (e.g.,
photovoltaic and wind resources); remotely located energy If the nonlinear vector valued function is linearized
storage devices (e.g., substation batteries [40]; flywheels [41]; around a presumed operating point (e.g., ), is
superconducting magnetic energy storage devices [42]). The replaced by where now denotes a constant
next generation of distribution systems would likely draw process matrix. It is a simple matter to show the linear unbiased
heavily on electronic controls for energy flow including various estimate of , which is the minimum square fit of the right and
configurations of ac/dc/ac converters and ac/ac converters. left hand sides of (3), is given by
The challenges in this area include the design of cost effective
controllers that are capable of operating in distribution system (4)
environments. Several alternatives include the following.
• Hybrid configurations in which the most effective loads where refers to the pseudoinverse of a matrix. A test on
are electronically managed while other loads are fed from the 2-norm of the residual determines whether it
conventional magnetic transformers. is necessary to perform the linearization again around the newly
• The use of magnetic amplifier-type technologies to control estimated value of . The process proceeds iteratively to obtain
some loads and perhaps some circuits by controlling mag- the final estimate of .
netic saturation in control windings. The traditional transmission application of state estimation
• The use of dedicated, asynchronous circuits, possibly of utilizes a state vector which is entirely real and consists of
various “power frequencies,” for example 40 kHz low elements that are mainly polar form magnitudes and polar form
current high-voltage distribution for electronic lighting angles of system voltages. However, it is possible to reformu-
in large buildings. Consideration might be given to low late the state estimation problem in terms mainly of synchro-
frequency ac for modest distance circuits connecting wind nized measurements (i.e., phasor quantities). In the synchro-
resources from induction generators to the load and to the nized phasor formulation, the rectangular form is used for el-
ac distribution system. ements in . If the state vector consists of rectangular form
• The design of hybrid fault interruption devices, including complex entries, consider the partitioning of the complex vector
micromechanical electric machine technologies [33], a into (the real part of ), and (the imaginary part of ),
232 IEEE TRANSACTIONS ON SMART GRID, VOL. 1, NO. 3, DECEMBER 2010

and the same notation for the partitioned measurement vector. “next generation” topics in distribution engineering. Perhaps the
Then the residual of a suitably formulated form of (2) is greatest single impetus in education and training has come from
the Congress of the United States which dedicated 16.4 billion
(5) U.S. dollars in 2009 to energy efficiency and renewable energy
topics in a range of educational venues. A 2005–2006 IEEE
Then it is possible to avoid the nonanalytic as follows: Power and Energy Society survey [47] showed that:
• 49.1 percent of 114 undergraduate and graduate educa-
tional programs in Canada and the U.S. have a dedicated
course in transmission and distribution engineering;
(6) • 61.5 percent of the programs polled have renewable re-
sources in the curriculum.
Minimizing the 2-norm of and , gives the same result as (4) The advent of Internet-based courses and other cyber tools for
despite the fact that is now a complex vector. Note that this training indicate that advanced topics in power engineering can
is a consequence of minimizing the independent real and imag- readily be presented to geographically diverse students, even
inary parts of the complex residual vector . The simultaneous those who are practicing engineers [48]. Despite these positive
minimization is equivalent to estimating and . This means aspects, there appears to be a shortage of power engineers in
that the efficient software that evaluates the pseudoinverse of a general [49], and distribution engineers in particular. And there
complex matrix may appropriately be applied to the formula- appears to be a shortage instructors and university professors in
tion of the state estimation problem is an estimate of a complex distribution engineering.
vector . Moreover, the formulation shown is completely linear
if the measurements are entirely voltage and current phasor V. DISTRIBUTION ENGINEERING ECONOMICS
synchronized measurements. In the absence of change in circuit
configuration, needs to be calculated only once. This ap- A. The Distribution Class LMP Signal
proach has been tested in simulation and found to be suitable Much of the electric power industry has been cost/benefit
for distribution state estimation. The condition number [46] of driven for many years. It has been estimated that the investment
the process matrix is found to be well below that of conventional in automating and bringing the distribution system in the United
transmission state estimation. States to the most recent state of technology requires 127 billion
U.S. dollars (as compared to $38 billion for the transmission
D. Consequences of Embedded Renewable Generation system) [50]. And the benefit of this investment is estimated to
The integration of integrate renewable generation resources be between 4:1 and 5:1 in distribution engineering. These data,
into power systems in general is motivated by the general public, even if inaccurate, suggest that there shall be a considerable
needs to reduce CO production, and reduction in demands from focus on distribution engineering to realize benefits of future
the transmission system. The distribution system offers such investments in the overall power system. Drivers in this area
a venue. Wind and solar resources have uncertain, stochastic include the public impetus for disbursed generation that does
availability, and without energy storage, these resources mainly not produce greenhouse gasses, allocation of costs where they
impact energy demand at the substation. The distribution system are warranted, maintaining high levels of reliability appropriate
is likely to be required to serve full load levels with a diversity for a digital society, and compliance with a view to moderniza-
factor applied. The area of high penetration of renewable re- tion. In transmission systems, one signal that effectuates some
sources in distribution systems is a subject of high research im- of these drivers is the locational marginal price (LMP) which is
portance. References [53] and [54] are a small sampling of the defined as the cost to deliver the next megawatt hour to a given
literature of the area. transmission bus [51], [52]. The LMP is, in effect, a low-pass
filtered signal of price related to the generation, processing, and
E. Communications Needs transmission of energy. In transmission systems, the LMP is
found by a multidimensional minimization of the cost of the
The automation of the distribution system has concomitant
next megawatt-hour delivered to a given system bus. The con-
requirements for sensors, communications, and computation.
cept of an LMP may be translated to a distribution system milieu
The most widely accepted communication protocol for power
by adopting the basic concept of cost of energy at the substation,
distribution is the International Electrotechnical Commission
i.e., the transmission LMP, and inclusion of distribution objec-
(IEC) Standard 61850 [55]. This standard is communication
tives and costs. Such a reformulated LMP is the Distribution
architecture suitable for TCP/IP networks and substation
LMP (D-LMP) which may be defined as
local area networks using a high-speed ( 1/4 cycle) switched
Ethernet. D-LMP LMP
sources
F. Training/Education
Much of the development of distribution engineering lies in
lines
the hands of those who practice this specialty. This implies that
training and education, in all forms including undergraduate, (7)
graduate, and continuing education need to include innovative renew gen lines
HEYDT: THE NEXT GENERATION OF POWER DISTRIBUTION SYSTEMS 233

Fig. 6. The use of D-LMP as a control signal in distribution systems.

In this formulation, the D-LMP differs from its transmission TABLE II

counterpart in that added terms are inserted to capture concep- THE NEXT GENERATION OF DISTRIBUTION SYSTEM CONTROLS
tualized objectives such as renewable resource encouragement.
The first term of the D-LMP is the transmission LMP at the
supply substation and denotes the generation participation
factor at connection bus . In the case of multiple grid tie points,
the participation factors are used to weight the several trans-
mission/subtransmission bus LMPs according to the energy (per
hour) obtained from that substation. The second term in (7) is
a term that captures the cost of losses in the distribution system
in each line ( denotes line resistance). This term is uses
to obtain the incremental line active power
lost. The third term of (7) is a term that captures the objective
of encouragement of the use of renewable resources at the given
load point. The last term in (7) relates to the circuit loading rep-
resented as a fraction of the circuit rating , and this is a
cost associated with reduction in congestion in a networked dis-
tribution system. The in (7) are weights. It is evident that
selected terms 2 through 4 could be replaced or redefined (or
zeroed) as desired in achieving a given set of distribution en- which allows the distribution system operator to move loads de-
gineering objectives. If the system is not networked, term four pending on operational needs. Fig. 6 shows the basic concept of
would be dropped. D-LMP as a control signal: note that calculation of the D-LMP
It is important to mention that (7) is not the only way to de- is readily done locally or centrally, and the calculation requires
fine a D-LMP. The formulation shown here incorporates some system-wide measurements or estimates (e.g., from a distribu-
operating objectives in a way reduced to a “price signal.” Other tion class state estimator).
objectives could be incorporated. Or, the D-LMP could reflect
only the cost of the next kilowatt hour delivered. The definition B. System Growth and Voltage Levels
of D-LMP would depend on the application envisioned.
The D-LMP is potentially a useful signal for control purposes. Distribution system growth is a key element of distribution
Like the transmission LMP analog, the D-LMP is a low pass company engineering efforts. The primary voltage level has
filtered energy “cost” signal where the term “cost” is broadly risen from the 5.0 kV class in the 1970s, to the common 15 kV
defined as a weighted signal to achieve given objectives. The class today, and this appears to be progressing to 35 kV at
D-LMP at each load bus may be used for control of local bus this time. The economic slowdown of 2008–11 may retard
loads such as those listed as bus load controls in Table II. In this effort (e.g., distribution system growth is estimated in the
the control of bus loads, an electronic controller is used to vary United States at 1% per year in [27]). However, the 330 billion
power (e.g., supplied to—or taken from an energy storage de- US dollar investment in distribution systems (2004–2024) and
vice, or in an energy management application). The D-LMP the impetus to renovate distribution systems and incorporate
signal may be used to vary set points in these electronic con- renewable resources in them suggest that the slow growth is
trollers and gains in the automatic control feedback loops. The a brief respite for system growth. Widely varying estimates
time resolution used in testing of the D-LMP as a control signal of growth due to electric vehicle loads may disagree on exact
indicates that a suitable range is s. This range figures, but they seem to agree that this load type is inevitable.
would correspond to similar calculations in the transmission
system. The component costs (i.e., the constants) in (7) are se- C. Finance Infrastructures
lected to reflect accurate values of the transmission LMPs, and A key element in distribution system design is cost man-
estimated costs of losses and congestion. It is also possible to agement. The present configurations of distribution companies
augment the D-LMP with a “distribution operator assist” term often supports innovative rate structures and cost allocation
234 IEEE TRANSACTIONS ON SMART GRID, VOL. 1, NO. 3, DECEMBER 2010

plans (e.g., energy management systems, time of day me- [16] Power quality improvement in hot rolling mill by means of SVC.
tering), but it is clear that as renewable resources penetrate Västerås, Sweden, ABB Power Systems AB, Application note A02-
0173 E.
into distribution engineering, and as governmental directives [17] C. Oliveira, N. Kagan, A. Meffe, S. Jonathan, S. Caparroz, and J.
require various energy saving and CO reduction measures, Cavaretti, “A new method for the computation of technical losses in
further advances in financial infrastructures are possible. One electrical power distribution systems,” in Proc. 16th Int. Conf. Exhib.
Elect. Distrib., Jun. 2001, vol. 5, pp. 1–5, IEE Pub. 482, Part 1.
example is the use of the D-LMP described above. Another [18] Technology Options 2003. Washington, DC, 2005, Anonymous, U.S.
is the optimal management of several objectives—that is, Climate Change Technology Program, updated.
multiobjective optimization. There is no doubt that in the long [19] R. Saint, “DOE distribution transformer efficiency standards,” pre-
sented at the IEEE Rural Elect. Power Conf., Charleston, SC, Apr. 29,
run, the management of cost, environment, fuel, and reliability
2008.
will be a challenge. [20] C. A. Dortolina and R. Nadira, “Emerging trends in engineering and
technology,” IEEE Trans. Power Syst., vol. 20, no. 2, pp. 1119–1125,
May 2005.
VI. CONCLUSION [21] S. Karandikar and A. Ghatol, “Combating overload in distribution
system to control power loss and increase reliability,” in Proc. 1st Int.
Several trends in distribution engineering have been noted, Conf. Emerging Trends Eng. Technol., Jul. 2008, pp. 917–921.
and several recommendations have been proposed in this paper [22] A. L. Jones, B. E. Smith, and D. J. Ward, “Considerations for higher
for the use of sensory information to automate and upgrade voltage distribution,” IEEE Trans. Power Del., vol. 7, no. 2, pp.
782–788, Apr. 1992.
power distribution systems. This is viewed as being consistent [23] M. Al-Muhaini, G. Heydt, and A. Huynh, “The reliability of power
with efforts in the smart grid philosophy. The sensory infor- distribution systems as calculated using system theoretic concepts,” in
mation includes the use of a distribution class locational mar- Proc. IEEE Power Eng. Soc. Gen. Meet., Minneapolis, MN, Jul. 2010,
pp. 1–8.
ginal price to drive energy management related controls; and a [24] G. Heydt, “Improving distribution reliability (the “N9 Problem”) by
distribution class state estimation procedure has been described the addition of primary feeders,” IEEE Trans. Power Del., vol. 19, no.
based on a noniterative algorithm that uses synchronized voltage 1, pp. 434–435, Jan. 2004.
[25] R. Pérez, G. Heydt, N. Jack, and A. Castelhano, “Optimal restoration of
and current measurements. Electronic solutions for power flow distribution systems using dynamic programming,” IEEE Trans. Power
control are described. It appears that hybrid electronic-conven- Del., vol. 23, no. 3, pp. 1589–1596, Jul. 2008.
tional distribution engineering approaches offer significant ad- [26] Interconnection of distributed energy resources in secondary distribu-
vantages. The reliability of distribution systems is proposed to tion network systems, Elect. Power Res. Inst., Palo Alto, CA, EPRI
White Paper 1012922, Dec. 2005.
be enhanced through automated restoration of service after an [27] G. Heydt, Electric Power Quality. Scottsdale, AZ: Stars in a Circle
outage. The upgrading of circuit voltages to the 35 kV class is Publ., 1991.
recommended to improve reliability and reduce losses. [28] W. Rosehart, A. Schellenberg, and C. Roman, “New tools for power
system dynamic performance management,” in Proc. IEEE Power Eng.
Soc. Gen. Meet., Jun. 2006, pp. 1–7.
REFERENCES [29] I. Das and J. Dennis, “Normal boundary intersection: A new method
for obtaining Pareto optimal points in multicriteria optimization prob-
[1] Electrical Transmission and Distribution Reference Book. E. Pitts- lems,” SIAM J. Optim., vol. 8, no. 3, pp. 631–657, 1998.
burgh, PA: Westinghouse Electr. Corp., 1964. [30] J. Kyei, R. Ayyanar, G. Heydt, R. Thallam, and J. Blevins, “The design
[2] W. H. Kersting, Distribution System Modeling and Analysis. Boca of power acceptability curves,” IEEE Trans. Power Del., vol. 17, no. 3,
Raton, FL: CRC, 2006. pp. 828–833, Jul. 2002.
[3] T. A. Short, Electric Power Distribution Handbook. Boca Raton, FL: [31] G. A. Putrus, M. M. R. Ahmed, I. Ran, K. R. Chu, and T. Funabashi,
CRC, 2003. “Integration of fault current limiters in power distribution networks,”
[4] T. Gönen, Electric Power Distribution Engineering. Boca Raton, FL: in Proc. IEE Int. Conf. Develop. Power Syst. Protection, 2004, vol. 1,
CRC, 2007. pp. 311–314.
[5] J. Northcote-Green and R. G. Wilson, Control and Automation of Elec- [32] B. W. Lee, J. Sim, K. B. Park, and I. S. Oh, “Practical application issues
trical Power Distribution Systems. Boca Raton, FL: CRC, 2007. of superconducting fault current limiters for electric power systems,”
[6] Smart Grid. Washington, DC, Dec. 2007, 110th Congress of United IEEE Trans. Appl. Supercond., vol. 18, no. 2, pp. 620–623, Jun. 2008.
States, Title XIII, Energy Independence and Security Act of 2007. [33] G. G. Karady, G. T. Heydt, E. S. Gel, and N. F. Hubele, “The utiliza-
[7] T. F. Garrity, “Innovation and trends for future electric power systems,” tion of micromechanical devices in a power circuit breaker,” J. Power
in Proc. IEEE Power Syst. Conf. Expo., 2009, pp. 1–8. Compon. Syst., vol. 33, no. 10, pp. 1159–1174, Oct. 2005.
[8] Recommended Practice for Electric Power Distribution for Industrial [34] A. V. Mamishev, B. D. Russell, and C. L. Benner, “Analysis of high
Plants, Standard 141-1986, IEEE, Piscataway, NJ. impedance faults using fractal techniques,” IEEE Trans. Power Syst.,
[9] Recommended Practice for the Design of Reliable Industrial and Com- vol. 11, no. 1, pp. 435–440, Feb. 1996.
mercial Power Systems, Standard 493, 2007, IEEE, Piscataway NJ. [35] Standard for Ventilated Dry-Type Power Transformers, 501 kVA
[10] R. Pérez and G. T. Heydt, “Distribution system restoration via subgra- and Larger, Three-Phase, With High-Voltage 601 V to 34 500 V;
dient based Lagrangian relaxation,” IEEE Trans. Power Syst., vol. 23, Low-Voltage 208Y/120 V to 4160 V—General Requirements, Standard
no. 3, pp. 1162–1169, Aug. 2008. C57.12.51, 2008, IEEE, Piscataway NJ.
[11] M. S. Lee and J. T. Lim, “Restoration strategy for power distribu- [36] D. J. Ward, “Overvoltage protectors—A novel concept for dealing with
tion networks using optimal supervisory control,” in Proc. IEE Gener., overbuilt distribution circuits,” IEEE Trans. Power Del., vol. 25, no. 3,
Transm., Distrib., May 2004, vol. 151, no. 3, pp. 367–372. pp. 1971–1977, Jul. 2010.
[12] R. P. Fanning, “Implementation of networked primary and secondary [37] “10 CFR part 431 energy conservation program for commercial equip-
distribution systems for U. S. utilities,” in Proc. IEEE Power Energy ment: Distribution transformers energy conservation standards,” Fed-
Soc. Gen. Meet., 2003, vol. 4, pp. 2425–2429. eral Register Oct. 12, 2007, U.S. Dept. Energy.
[13] S. J. Dale, “Ship power system testing and simulation,” in Proc. IEEE [38] S. Bhattacharya, T. Zhao, G. Wang, S. Dutta, S. Baek, Y. Du, B.
Elect. Ship Technol. Symp., 2005, pp. 202–205. Parkhideh, X. Zhou, and A. Q. Huang, “Design and development of
[14] S. A. Lewis and T. L. Baldwin, “Distribution load flow methods for generation-I silicon based solid state transformer,” in Proc. 25th Annu.
shipboard power systems,” in Conf. Rec. 38th IEEE Ind. Appl. Soc. IEEE Appl. Power Electron. Conf., 2010, pp. 1666–1673.
Annu. Meet., 2003, vol. 2, pp. 812–819. [39] J. De La Ree, V. Centeño, and A. G. Phadke, “Synchronized phasor
[15] IEEE Standard for Shunt Power Capacitors, Standard C18-1992, 1992, measurement applications in power systems,” IEEE Trans. Smart Grid,
IEEE, Piscataway NJ. vol. 1, no. 1, pp. 20–27, Jun. 2010.
HEYDT: THE NEXT GENERATION OF POWER DISTRIBUTION SYSTEMS 235

[40] F. A. Chacra, P. Bastard, G. Fleury, and R. Clavreul, “Impact of energy [52] C. Xu and T. Overbye, “An energy reference bus independent LMP
storage costs on economical performance in a distribution substation,” decomposition algorithm,” IEEE Trans. Power Syst., vol. 21, no. 3, pp.
IEEE Trans. Power Syst., vol. 20, no. 2, pp. 684–691, May 2005. 1041–1049, Aug. 2006.
[41] M. M. Flynn, P. McMullen, and O. Solis, “Saving energy using fly- [53] P. Siano, P. Chen, Z. Chen, and A. Piccolo, “Evaluating maximum
wheels,” IEEE Ind. Appl. Mag., vol. 14, no. 6, pp. 69–76, Nov./Dec. wind energy exploitation in active distribution networks,” Proc. IET
2008. Gener., Transm., Distrib., vol. 4, no. 5, pp. 598–608, 2010.
[42] K. Tam and P. Kumar, “Impact of superconductive-magnetic energy [54] A. Keane, Q. Zhou, J. Bialek, and M. O’Malley, “Planning and op-
storage on electric power transmission,” IEEE Trans. Energy Convers., erating non-firm distributed generation,” Proc. IET Renewable Power
vol. 5, no. 3, pp. 501–509, Sep. 1990. Gener., vol. 3, no. 4, pp. 455–464, 2009.
[43] A. G. Phadke, J. S. Thorp, and K. Karimi, “State estimation with phasor [55] IEC 61850, Communication Networks and Systems in Substations.
measurements,” IEEE Trans. Power Syst., vol. PWRD-1, no. 1, pp. Geneva, Mar. 2010, Int. Electrotech. Commission (IEC).
233–241, Feb. 1986. [56] S. Rahman and G. Shrestha, “An investigation into the impact of elec-
[44] Standard for Synchrophasors for Power Systems, Std. C37.118—2005, tric vehicle load on the electric utility system,” IEEE Trans. Power Del.,
IEEE, Piscataway NJ. vol. 8, no. 2, pp. 591–597, 1993.
[45] A. Abur and A. G. Expósito, Power System State Estimation: Theory [57] J. Taylor, A. Maitra, M. Alexander, D. Brooks, and M. Duvall, “Eval-
and Implementation. New York: Marcel Dekker, 2004. uation of the impact of plug-in electric vehicle loading in distribution
[46] M. J. Rice and G. T. Heydt, “The measurement outage table and state system operations,” in Proc. IEEE Power Energy Soc. Gen. Meet., Cal-
estimation,” IEEE Trans. Power Syst., vol. 23, no. 2, pp. 353–360, May gary, AB, 2009, pp. 1–6.
2008.
[47] “Electric power engineering education resources 2005–2006,” IEEE
Trans. Power Syst., vol. 23, no. 1, pp. 1–24, Feb. 2008, IEEE Power
Engineering Society, Power Engineering Education Committee Task
Force.
[48] K. Nigim, G. Heydt, and J. Palais, “E-learning opportunities for elec-
tric power engineers,” IEEE Trans. Power Syst., vol. 22, no. 3, pp.
1382–1383, Aug. 2007. Gerald Thomas Heydt (SM’62–M’64–SM’80–
[49] W. Reder, A. Bose, A. Flueck, M. Lauby, D. Niebur, A. Randazzo, D. F’91–LF’08) received the Ph.D. degree in electrical
Ray, G. Reed, P. Sauer, and F. Wayno, “Engineering the future,” IEEE engineering from Purdue University, West Lafayette,
Power Energy Mag., vol. 8, no. 4, pp. 27–35, Jul./Aug. 2010. IN, in 1970.
[50] Power delivery system of the future: A preliminary estimate of costs He is currently the site director of a power engi-
and benefits Elect. Power Res. Inst., Palo Alto, CA, EPRI Rep. neering center program at Arizona State University,
1011001, Jul. 2004. Tempe, where he is a Regents’ Professor.
[51] E. Litvinov, “Design and operation of the locational marginal prices Dr. Heydt was awarded the Richard Harold
based electricity markets,” Proc. IET Gener., Transm., Distrib., vol. 4, Kaufmann award for excellence in distribution
no. 2, pp. 315–323, 2010. engineering in 2010.