IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO.

2, MAY 2013 1493

State-of-the-Art Techniques and Challenges Ahead

for Distributed Generation Planning and Optimization
Task Force on Distributed Generation Planning and Optimization
Andrew Keane, Member, IEEE, Luis F. Ochoa, Senior Member, IEEE, Carmen L. T. Borges, Senior Member, IEEE,
Graham W. Ault, Member, IEEE, Arturo D. Alarcon-Rodriguez, Robert A. F. Currie, Fabrizio Pilo, Member, IEEE,
Chris Dent, Member, IEEE, and Gareth P. Harrison, Member, IEEE

Abstract—It is difficult to estimate how much distributed gen- more cost-effective for society in the long term—is potentially
eration (DG) capacity will be connected to distribution systems left behind.
in the coming years; however, it is certain that increasing pene- Although it is an imperative for many countries to aggres-
tration levels require robust tools that help assess the capabilities
and requirements of the networks in order to produce the best sively promote the connection of low-carbon and efficient
planning and control strategies. The work of this Task Force is generation technologies, different government policies and
focused on the numerous strategies and methods that have been regulatory frameworks have resulted in different technical
developed in recent years to address DG integration and planning. and economic drivers for DNOs towards the connection of
This paper contains a critical review of the work in this field. Al- DG. Depending on the incentives in place, DNOs not tied to
though there have been numerous publications in this area, wide-
spread implementation of the methods has not taken place. The unbundling rules might be able to increase network reliability
barriers to implementation of the advanced techniques are out- or postpone reinforcements by investing in and operating ade-
lined, highlighting why network operators have been slow to pick quately sited and sized DG plants. On the other hand, unbundled
up on the research to date. Furthermore, key challenges ahead DNOs capable of determining optimal locational connection
which remain to be tackled are also described, many of which have charges might steer the deployment of DG in areas that could
come into clear focus with the current drive towards smarter dis-
tribution networks. potentially reduce their energy losses (or at least not worsen
them). From a purely operational perspective, DG plants could
Index Terms—AC optimal power flow, active network manage-
also be encouraged to provide reactive power support if that is
ment, distributed generation, distribution networks, linear pro-
gramming, multi-objective programming, wind power generation. of concern for the DNO or the regional transmission network.
This variety of cases where DG technologies can play a major
role in distribution networks has in the last decade encouraged
I. INTRODUCTION researchers and industrialists around the world to investigate
the corresponding planning and operational aspects.
F OR distribution network operators (DNOs) the challenges
posed by high penetrations of distributed generation (DG)
are numerous. In fully liberalized electricity markets (e.g., Eu-
Whatever the particular driver for a DNO, e.g., to allow the
connection of more DG capacity, to reduce energy losses, or to
ropean Union), planning the siting and sizing of DG plants is, increase network reliability, these DG planning tools must take
in many respects, not possible. Due to unbundling rules, DNOs into account essential network constraints such as voltage and
cannot invest in generation facilities and are meant to provide thermal limits. The inherent variability of demand and renew-
DG owners with cost-effective connection means, irrespective able generation (e.g., wind power) is an aspect that should also
of the technology or geographical location. In this context, un- be considered. In addition, characteristics of actively managed
certainties due to, for instance, planning consents or financial networks (as opposed to the current “fit and forget” approach),
support surrounding DG investments pose DNOs with major where control schemes employing real-time control and com-
challenges as to what, where and when to reinforce the system in munication systems allow more effective management of dif-
order to deliver timely connections without the risk of stranded ferent network participants, including DG plants, voltage regu-
assets. This lack of certainty and planning coordination trans- lation devices, storage and demand, need also to be accounted
lates into distribution network operators often connecting DG for.
plants in a “fit and forget”, case-by-case manner where only This paper is structured as follows: First, Section II offers
traditional reinforcements (e.g., new lines or transformers) are a comprehensive review of the literature in the field, detailing
carried out. Thus, any sophisticated solution—albeit potentially the various methods and state-of-the-art techniques employed to
date. Section III gives a description of a number of the barriers
to implementation. Finally, Section IV describes the key future
Manuscript received March 06, 2012; revised July 06, 2012; accepted August
14, 2012. Date of publication September 19, 2012; date of current version April
challenges in this area, with conclusions given in Section V.
18, 2013. Paper no. TPWRS-00232-2012.
The authors are members of the IEEE PES Task Force on DG Planning and
Optimization, organized under Distribution System Analysis Subcommittee,
II. STATE-OF-THE-ART TECHNIQUES
and co-sponsored by the Power System Planning and Implementation (PSPI)
The optimal accommodation of conventional and renewable
and Power System Analysis, Computing and Economics (PSACE) committees
(e-mail: [email protected]; [email protected]). DG plants has been approached in the literature from different
Digital Object Identifier 10.1109/TPWRS.2012.2214406 angles but primarily taking account of technical and economic

0885-8950/$31.00 © 2012 IEEE

1494 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO. 2, MAY 2013

issues. This section presents a critical review of the various network. Such an approach might be useful if discrete values
methods and state-of-the-art techniques employed to date. (i.e., specific DG capacities) are preferable. However, the ac-
tual benefit brought by exhaustive analyses is that it is pos-
A. Analytical Analysis sible to cater for a number of technical issues and constraints.
If only a given demand-generation snapshot scenario is taken Indeed, with this more direct approach the objective function
into account, a specific technical aspect (or objective function) can be the combination of parameters or indices that represent
can be formulated analytically in such a way that it is possible different technical and non-technical aspects, although it will
to find the most beneficial DG capacity (or power injection) be very time consuming. This methodology was adopted in [3]
through a simplified set of equations and procedures. For in- and [4] where the siting and sizing of DG plants was inves-
stance, consider the simplified voltage drop formula (losses are tigated considering technical impacts such as power and en-
neglected) for a given line between nodes a and b: ergy losses, voltage rise, and short-circuit levels, which were
combined into a weighted-sum objective function. In [5], eco-
(1) nomic and environmental impact indices were also incorpo-
rated. Although a relatively straightforward technique, tuning
where is the distribution transformer or substation, and
the corresponding weighting factors to obtain a composite index
the resistance and reactance of the line, and and the active
becomes a non-trivial task that if not appropriately performed
and reactive components of the load connected to . Connection
can significantly bias the results. Moreover, while exhaustive
of a DG plant at with active and reactive power
analyses applied to a single connection is evaluated for a specific
output alters (2)
demand/generation scenario is not necessarily computationally
(2) intensive; this is not the case when multiple connections and
the variability of demand and generation are accounted for, in-
Equation (2) can be used to determine an “optimal” nominal ca- creasing considerably the computational burden of the exhaus-
pacity (MW) and power factor (or Mvar) that either minimizes tive analysis. However, it is important to highlight that the use
the voltage drop or avoids voltage rise beyond the statutory of state-of-the-art distribution analysis software packages such
limit. If a similar analytical analysis is extended to a network as OpenDSS [6] might prove a robust, fast alternative for ex-
and all possible locations for the potential DG development are haustively exploring extremely large search spaces. Moreover,
assessed, then a quick overview of “best” locations and sizes some metaheuristic optimization techniques (Section II-E) can
can be produced, for the snapshot scenario considered. efficiently explore the search space, reducing the computational
This type of analysis, focusing particularly on power losses, time to locate solutions.
was used in [1] and [2]. However, while power losses can
be studied in passive networks considering peak load sce- C. Linear Programming
narios—as is traditionally done—distribution networks with Linear programming (LP) has also been employed to address
DG plants require the assessment of energy losses. Not only the capacity allocation and energy optimization issues. Funda-
is this the actual yardstick used by DNOs but the inherent mentally, the use of linear programming entails a linearization
variable nature of demand and (renewable) generation neces- of the power flow or the linearization of the results from an
sitates it. Additionally, by neglecting other demand-generation ac power flow. It has been demonstrated through simulation
scenarios, technical issues that might appear otherwise, such that the resulting approximation inevitably introduces an error,
as voltage rise or thermal overloads, will not be accounted but not a significant one in the context of discrete turbine sizes
for since the analytical formulation only caters for a single [7], [8].
technical aspect (although they can be included to an extent In [9], a linear programming formulation of optimal power
in the corresponding solution procedure). Another limitation flow (OPF) is employed to assess the control of multiple DG
is that only a single DG plant can be evaluated at a time, plants. The objective employed is to minimize the annual active
requiring a sequential procedure if multiple connections are generation curtailment cost. The results presented illustrate the
needed. This separate evaluation of multiple DG plants might relative merits of tap changer and active and reactive power con-
result in the ’sterilization of capacity’ wherein inappropriately trol. In [10] ac power flow is employed to calculate linearized
located and/or sized plant prevents connection of larger plant sensitivity factors. The sensitivities are employed to charac-
elsewhere. The incorporation of operational solutions such as terize a range of constraints, such as voltage, thermal and short-
coordinated voltage control or generation curtailment cannot be circuit limits. The method is formulated as a linear program
done either. Consequently, although analytical approaches are and solved with the objective of maximizing the capacity of
straightforward alternatives to assess DG siting and sizing, care DG, subject to typical network constraints and taking account
must be taken as the results are only indicative and scenario of N-1 configurations. In [11] LP is applied to the question of
limited. non-firm DG access to the network. The objective employed is
to maximize the energy harvested from a section of network
B. Exhaustive Analysis by optimizing the allocation of DG with voltage constraints re-
A single technical issue, such as voltage rise or power losses, moved. The operation of DG has also been considered in the
can also be approached by exhaustively exploring the entire (or literature again employing ac load flow sensitivities to optimize
most of the) search space corresponding to the locations and the allocation of curtailment among adjacent wind farms [12].
sizes of DG plants that could be connected to a distribution An advantage of LP is that it offers significant potential for
KEANE et al.: STATE-OF-THE-ART TECHNIQUES AND CHALLENGES AHEAD FOR DISTRIBUTED GENERATION PLANNING AND OPTIMIZATION 1495

development of operational methods and is a robust optimiza- the maximum DG capacity able to be connected to a given
tion method. However, from a planning perspective, ac optimal network. More complex problems resulting from other network
power flow approaches would seem to be a more rigorous means constraints commonly overlooked by DG studies, such as N-1
of optimization at this stage. security, voltage step change, and fault levels are also viable
within this approach [8], [18], [19]. As in other optimization
D. AC Optimal Power Flow techniques, the same core framework can also be used to
The well-known ac OPF [13] has traditionally been used for investigate different objective functions, such as, for instance,
economic dispatch, and is widely acknowledged by the elec- the minimization of energy losses [20].
tricity industry as a powerful analysis tool. The ac OPF is a A number of solution methods can be adopted to solve the ac
nonlinear programming (NLP) problem, for which many solu- OPF problem: from special linear programming formulations to
tion methods exist including some which are highly specialized branch and bound techniques. Commercial solvers specialized
to OPF problems. The ac OPF formulation can be adapted to for NLP problems include CONOPT, that uses a generalized
have different objectives and constraints according to the study reduced gradient, and, KNITRO, that uses interior points. Al-
being carried out. For example, consider the minimization of though no practical method exists which can guarantee to find
power losses: the global optimum of a non-convex NLP, local optima can be
found in most cases. A NLP-formulated ac OPF will, of course,
(3) not cater for integer variables such as tap positions or discrete
values for DG capacities. It is possible to consider a mixed in-
where and are the active power injections at each end teger NLP approach, however, but this could potentially restrict
node (denoted 1 and 2) of branch . The difference between the the size of the problem depending on the capabilities of the so-
net injections at each end defines the line power losses. This lution method used with little benefit in the context of planning
objective would be subject to the standard Kirchhoff voltage law decisions. A notable example of this approach is in [21].
expressions, as well as a range of constraints that might include, The use of multi-periods to mimic the variability of genera-
for instance, bus voltage and line thermal limits. The Kirchhoff tion and demand results in a much larger problem in terms of the
current law (active and reactive nodal balance), however, will number of variables and constraints. In general this translates
need to be adapted to cater for the injection of potential DG into longer processing times, that depending on the size of the
plants: problem it could be intractable. In such a case, the problem has
to be scaled down by (mainly) reducing the number of periods.
(4) Methods for doing this include coincidence matching [17], “typ-
ical periods” [22] or clustering.
(5) Finally, it is important to highlight that given the nature of
this classical optimization technique, the problem has to be for-
mulated in a “closed” manner. While it is possible to do so with
where are the total power injections into lines at bus certain technical (and non-technical) aspects, for instance with
; and are the active and reactive demands at the same fault level calculation [18], this places a significant limitation as
bus. From the set of generation units, , the power injections, to what can be taken account of.
, of those connected to bus are also included. Thus, this
relatively simple formulation would lead to “optimal” siting and E. Metaheuristics
sizing of DG plants in a way that power losses are minimized A metaheuristic method is defined as an iterative generation
subject to the considered constraints. process which guides a subordinate heuristic by combining in-
Indeed the ac OPF has been used to tackle many of the same telligently different concepts for exploring and exploiting the
problems described in the LP section above. This alternative search space. Learning strategies are used to structure informa-
use of an OPF-like method for the (power) loss minimization tion in order to efficiently find near-optimal solutions. Complex
problem was reported in [14]. A similar formulation but with and “decoupled” technical and non-technical issues involved
the objective of maximizing DG capacity across multiple sites in power systems optimization problems can be easily mod-
has also been adopted in [15] and [16]. However, in these three eled and included in the optimization process. Metaheuristics
OPF-based approaches, only extreme cases of peak or minimum do not require the “closed” formulation of the different aspects
demand and passive operation of the network were considered. being addressed as is necessary in classical optimization. Meta-
The flexibility provided by a tailored ac OPF makes it heuristic algorithms can also cater for mixed integer problems
possible to extend the analysis to cater not only for voltage that are common in power system optimization problems.
and thermal limits but also for a number of complex aspects. With metaheuristics, the objective function can be of any type
It can incorporate multiple periods to deal with the variability and take into consideration different objectives. This character-
and coincidence of demand and renewable generation. Ad- istic leads to multi-objective applications that are better suited
vanced control strategies such as coordinated voltage control, to describe the complexity of the new distribution businesses
adaptive power factor and generation curtailment can also [23]–[27] (see Section II-F). On the other hand, all metaheuristic
be incorporated to evaluate potential benefits. The approach algorithms require a careful tuning of optimization parameters
proposed in [17] embeds these characteristics to determine that are essential for finding a good solution without excessive
1496 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO. 2, MAY 2013

computation time. The attention is then moved from the mathe- if the value of the objective function is greater than that of the
matical formalization of both objective function and constraints current solution. To prevent cycling, some moves are marked
to the algorithm parameters, which should allow a compromise “tabu” for a number of iterations; the length of the tabu list, the
between quality of solutions and computing time. The actual tabu-tenure, fixed or variable, guides the optimization.
challenge when using these techniques is then the tuning of the The SA is an algorithm that combines combinatorial
parameters that guide the optimization. Indeed, care should be search with a very simple metaheuristic that follows the
taken to avoid premature or slow convergence, particularly in cooling process of materials. Following an appropriate cooling
large scale applications. schedule, the SA has the potential to avoid local minima and
This leads to probably the most discussed disadvantage of converges to the global minimum in a reasonable computing
metaheuristics: their inability to find the global optimum. In- time. The parameters to tune are the annealing temperature, the
deed, they are very likely to find a reasonable solution, though number of iterations at constant temperature and the cooling
there is no guarantee of exactly how good this is. Multiple runs strategy. SA annealing has been used for multi-objective op-
are often used to counter this. Metaheuristic algorithms allow timization to minimize energy losses, polluting emissions and
the planning engineer to find not only a single optimum point, contingencies. In [34] the authors proved that SA performed
but a family of near-optimum planning alternatives. This feature better than GA and TS on the IEEE 30-bus test system, but the
of metaheuristics is particularly useful in DG allocation because comparison is difficult to be accepted because neither the GA
the DNO generally has little or no control on the DG integration nor TS were optimized as the SA was.
and different planning alternatives can be necessary to face un- PSO makes use of a velocity vector to update the current posi-
certainties and minimize risks. tion of each particle in the swarm. The position of each particle
is updated based on the social behavior that a population of indi-
F. Metaheuristic Algorithms viduals, the swarm in the case of PSO, adapts to its environment
There are numerous metaheuristic algorithms: ant colony by returning to promising regions that were previously discov-
optimization (ACO), artificial bee colony optimization (ABC), ered. Recently PSO has been used for DG allocation [35]. Com-
tabu search (TS), particle swarm optimization (PSO), simulated parisons have been made between GA and other metaheuristic
annealing (SA) including genetic algorithms (GA). All these methods, and in some cases PSO converged faster than other al-
algorithms have been used to solve the problem of optimal gorithms finding out good quality solutions.
allocation of DG. ACO and ABC are based on the dynamic of the social insect
GA mimics the process of evolution. The most promising in- population. The interactions are executed via multitude of var-
dividuals have greater chances of transmitting their genes to off- ious chemical and/or physical signals (e.g., bee dancing during
spring. By so doing, the population, generation by generation, the food procurement, ants’ pheromone secretion, and perfor-
improves and, if the premature convergence is avoided, for in- mance of specific acts, which signal the other insects to start the
stance, with a random mutation, the algorithm converges. GA same actions). The final product of different actions and inter-
have been used by the first authors that pioneered the problem actions represents the behavior of social insect colony. ABC al-
of the optimal integration of distributed energy resources in the gorithms has been used for determining the optimal DG-unit’s
distribution system and since then it has been preferred to other size, power factor, and location in order to minimize the total
metaheuristic algorithms [28]–[31]. The reason of the success system real power loss [36], [37].
is that GA is intrinsically suited to solve location problems. The
coding of a solution can be very simple, a binary vector with G. Multi-Objective Programming
as many positions as the number of bus candidate to DG con- Some DG planning objectives are naturally conflicting; con-
nection. The classical GA operators (selection, crossover and sequently in some cases there is no single planning solution that
mutation) can be used simply and effectively with few or no will satisfy all stakeholders. A multi-objective problem with
changes. As in many metaheuristic algorithms, high values of conflicting objectives has no single solution, but a set of solu-
penalty factors can be added to the fitness function of those in- tions, known as the Pareto set. The multi-dimensional concept
dividuals that do not comply with the constraints. It should be of “dominance” is used to determine if one solution is better than
recognized that with simple rules of thumb the parameters of other solutions [38]. All of the non-dominated solutions consti-
GA (i.e., population size, crossover type, etc.) can be set quite tute the Pareto-set.
easily to achieve a good optimization tool. Fuzzy GA have been Multi-objective optimization problems are solved by two
proposed using fuzzy genetic algorithm approaches in order to fundamentally different groups of techniques. The first set
capture the multi objective nature of problems [32]. Genetic al- of techniques uses preference information and the iterative
gorithms have also been applied to consider optimal investment repetition of a single-objective optimization problem, usually
planning for DG in a market structure [33]. solved by GA. The most common techniques of this first group
TS is a metaheuristic that guides a heuristic method to expand are the weighted-sum method and the -constrained method.
its search beyond local optimality, with the systematic prohibi- In the former, all objectives are aggregated to produce a single
tion of some solutions to prevent cycling and to avoid the risk objective problem (similar to [3]–[5]; see Section II-B), this
of being trapped in local minima. New solutions are searched in method is then iteratively used to change the set of weights
the set of the points reachable with a suitable sequence of local to find the Pareto set. In the -constrained method, one of the
perturbations (neighborhood). One of the most important fea- objectives is optimized while the rest are kept as constraints
tures of TS is that a new configuration may be accepted even to find each one of the solutions of the Pareto set [26], [30],
KEANE et al.: STATE-OF-THE-ART TECHNIQUES AND CHALLENGES AHEAD FOR DISTRIBUTED GENERATION PLANNING AND OPTIMIZATION 1497

[39]. These methods are useful when there is strong a priori DG plants based on dispatchable and storable energy sources,
knowledge of the problem, or when a particular region of the such as biomass, can be more easily modeled since energy can
search space is explored. However, a large number of iterations be considered available in reliability studies. The only issue usu-
must be performed to find many solutions of the Pareto set, ally considered for unavailability of generation is the failure of
increasing vastly the computational requirements, especially the generating unit. This kind of DG tends to be more reliable.
when many objectives are being analyzed. On the other hand, the units based on variable and non-storable
Another group of techniques, known as multi-objective ge- energy sources, such as wind, small hydro and solar, require
netic algorithms (MOGA), have been proposed in recent years a more complex model in reliability studies, where the energy
to overcome the above mentioned deficiencies, and to provide availability also needs to be represented.
a “true” multi-objective approach [40]. Indeed, MOGA are able Stochastic models for renewable sources have been devel-
to find many solutions of the Pareto set at once. Two of the oped for wind generation [49], small hydro power plants [50],
most powerful multi-objective evolutionary algorithms are the solar generation [51], and biomass thermal generation [52]. In
Non-Sorting Genetic Algorithm—II (NSGA-II) [41] and the general, generation availability of all these sources is obtained
Strength Pareto Evolutionary Algorithm 2 (SPEA2) [32]. A key by the combination of the availability model of the primary en-
advantage of the use of MOGA is the opportunity of using com- ergy source and that of the generation unit.
plex objective evaluations in the formulation of the problem, Models for reliability evaluation of distribution systems with
which can include stochastic simulations, OPF analyses, and distributed renewable generation are based on three different
probabilistic approaches, to provide more realistic models for approaches: analytical methods [27], Monte Carlo simulations
dispatchable and non-dispatchable DG and storage [22], [33], [48] and hybrid models [53]. Analytical methods are applicable
[42]–[44]. MOGA also permit to analyze a variety of technical, to DG of dispatchable energy sources. The Monte Carlo simu-
economical and environmental objectives in a single optimiza- lation approach is adequate to represent DG of variable energy
tion framework [22], [43], [45]–[47]. Multi-objective methods sources and also to aggregate the load variation curve. The hy-
by their nature find compromise solutions rather than a single brid model aims to combine the advantages of the first two ap-
solution. This is an advantage in terms of providing insight but proaches in terms of computational efficiency and the represen-
a drawback in that it leaves it open to interpretation by different tation of energy availability uncertainty.
parties and the ultimate decision will rest with system planners
In a general sense, it can be said that DG enhances the relia-
and operators.
bility of distribution systems especially if islanded operation is
allowed. However, when the generation is based on variable en-
H. Probabilistic Analysis ergy sources, the benefit is reduced and can even be negligible if
Uncertainties related to DG are due to two main aspects: the not properly planned. The probabilistic analysis is able to prop-
variable nature of the primary energy source and the possible un- erly capture this effect and to provide a more realistic response
availability of the unit when it is required to generate. The com- of the impact on the distribution system than the deterministic
bination of these aspects may lead to generation deficit, which techniques.
can heavily compromise the security, reliability and quality of However, probabilistic approaches have two most discussed
power supply. The increase in the complexity of distribution disadvantages: the large amount of data required and the poten-
systems with DG requires the assessment of the random nature tial difficulty in interpreting the results and thus make decisions
of network failures and generation availability [48]. based on such results. The probabilistic reliability evaluation
In order to adequately address the uncertainties introduced by or system planning requires the adequacy analysis of several
DG integration to distribution systems, probabilistic methods system states or expansion alternatives that are performed by
can be applied for network planning and optimization. Besides, optimization methods [54]. Therefore, the use of efficient opti-
stochastic models of renewable resources must be developed in mization methods in probabilistic analysis is crucial to have ac-
order to represent the influence of the primary energy source ceptable computation time. An approach that is being explored
variability on generation availability. The impact of DG on the to reduce the computational burden is to use optimization tech-
reliability of distribution systems depends mainly on the oper- niques (classical or metaheuristics) for state space pruning in
ational mode and purpose of connection to the system along stochastic simulation [55].
with the energy source which drives it. For instance, DG could
mainly be used to supply power to a local load (e.g., indus- I. Summary
trial site or a house), exporting to the distribution network only The division of sections in the previous sections is along the
when there is excess capacity. Depending on commercial ar- lines of the computational methods employed. As is evident the
rangements, such a consumer will only pay (or will be paid) the problems and objectives of the methods cut across many of the
difference between the energy consumed and exported. In this computational methods. Table I summarizes the cross section
case, there is no benefit to system reliability (only to the con- of methods and objectives and their relationship based on the
sumer’s reliability). However, if DG operates in parallel with extant literature. It is evident that the siting and sizing of DG
the network, then new considerations must be introduced for has been a focus with objectives ranging from minimization of
reliability modeling. The simplest alternative is to model DG as losses to maximization of installed capacity to multi-objective
a negative power injection, which can have a positive impact on approaches taking account of a number of objective simultane-
reliability since it represents a reduction in demand. ously and examining the trade-off between them.
1498 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO. 2, MAY 2013

TABLE I network operators and DG developers the opportunities for and

SUMMARY OF TECHNIQUES AND OBJECTIVES level of network access. The treatment of uncertainty in DG
planning is dealt with in Section II-G but the uncertainty of the
operational characteristics of DG is assessed alongside the via-
bility of DG constraint management solution using a stochastic
programming approach in [64].
On the demand side, the use of new data sources in distribu-
tion and DG planning and optimization is both a challenge and
an opportunity. The wealth of data (and hopefully meaningful
information) available from smart meter roll out programs pro-
vides a challenge to network planners (e.g., aggregation issues
for domestic level data) but also a rich insight into the possibil-
ities for DG deployment or network access at domestic scales
upwards. It is anticipated that this will be an area of substantial
development in the coming years based on early publication of
results on this topic [65], [66]. The complete area of data cap-
ture, processing, modeling, estimation, forecasting as it relates
to studies of DG and DG network integration seems to be of in-
III. BARRIERS TO IMPLEMENTATION creasing importance.
There are numerous implementation challenges in integrating A further issue for optimal network planning with DG using
DG into power systems and these have been explored well at “active network management” approaches is proper represen-
a research level in recent years [56]–[58]. Many of these have tation of the secondary communications and control infrastruc-
been viable solutions and some of those have found their way ture. On the one hand the cost of this needs to be incorporated
into application in different regions around the globe. For ex- into the network planning formulation and on the other hand the
ample, one of the initial major moves to overcome DG con- impacts of these secondary systems (e.g., on reliability of net-
nection and integration barriers was updating grid connection work connection) need to be evaluated.
standards [59], [60]. Along with the DG technology and eco- This paper mainly deals with the planning issues of DG opti-
nomics issues, the network challenges for DG integration are in mization and planning in the medium or long term, but there is a
the “planning” and “operational control” areas (or time frames). case now being made for operational optimization of DG. In part
Two major barriers to implementation lie in the fact that DNOs this is due to the maturing of active network management ap-
are not used to using custom (if any) optimization codes and proaches and the requirement for DG network access to be max-
that some of the proposed techniques cannot consider a suffi- imized as far as possible. Some authors have proposed optimiza-
cient range of scenarios to tackle real world problems. tion approaches for DG control [67]–[70]. Some of the issues
Techniques for planning have been the focus of the discus- experienced with optimization approaches for DG operations
sion in the sections above and the DG optimal siting and sizing are scalability of approaches for many DG units, robustness in
problem (and proposed solutions) has been noted. In addition, finding a safe or feasible solution in control timescales, and the
there are challenges in adequately and economically planning possibility of hunting for solutions in a dynamic environment.
the sizing of distribution network circuits for DG connection. The DG integration domain has transitioned from one of
This is problematic since the techniques for managing the op- “learning by research” to “learning by doing” in the past few
eration of connected DG have a large role to play in assessing years. The advent of major stimulus or innovation funded smart
the required capability of the distribution network. Earlier work grid project with DG as an integral ingredient has brought
in this area simply attempted to sum up potential benefits and opportunities to develop, validate and demonstrate new ap-
costs of DG to the network operator to establish the case for proaches in DG integration. In the U.S. the ARRA [71], in the
DG with a simplified approach to representing the through year U.K. the Low Carbon Networks Fund [72], and similar schemes
operation [61]. elsewhere are at the stage of trialing very new approaches to
The DG planning optimization approaches described above DG integration. In most cases the emphasis on removing bar-
have some representations of the operational approach under riers to DG development is related to the low carbon emission
consideration (e.g., in the OPF formulation of DG curtailment characteristics of the DG technologies. The approaches being
in [17] and [43]) but fully representing each of the new ap- trialed include using energy storage, demand response, smart
proaches and their physical and operational characteristics and meter data, DG constraint management and other more active
constraints is a non-trivial task. A further complicating factor approaches to managing distribution networks [73].
is the uncertainty associated with these technologies (e.g., de- One issue that has emerged is the issue of DNO capability to
mand response, electric transportation, electric heating, energy adopt the more innovative solutions available. The complexity
storage) and where they will be connected and how they will of the planning and design process for networks with DG (and
be managed. There are already well-established outcomes from other new devices) and the integration of new operational ap-
studies on DG curtailment [17], [62], [63] to highlight DG ac- proaches has initially resulted in relatively slow progress on de-
cess opportunities (or headroom calculations as they are some- ployment projects. This highlights a major problem for state-of-
times known) and these prove useful in highlighting to power the-art DG planning and optimization approaches: “DNO adop-
KEANE et al.: STATE-OF-THE-ART TECHNIQUES AND CHALLENGES AHEAD FOR DISTRIBUTED GENERATION PLANNING AND OPTIMIZATION 1499

tion”. DNOs have a substantial capability challenge to embrace Low-voltage networks are now also becoming the focus of
whether receiving the results from more sophisticated planning attention of researchers and network operators. The potential
tools or developing models themselves or accepting for internal uptake of micro-generation, electric vehicles and other demand
use some of the tools and techniques emerging from research. side resources, coupled with developments in advanced me-
tering have led to new modeling efforts of these networks. Chal-
IV. CHALLENGES AHEAD lenges are present in the modeling of these networks alone, as
The move towards higher penetrations of DG is leading to historically they have not been modeled in detail by network op-
DG constituting a high percentage of the overall generation erators. Beyond the modeling challenge, lies the development
plant mix. In systems with a few interconnections, this scenario of optimization strategies and integration techniques for dis-
poses many challenges. For instance, in Ireland it is now ap- tributed energy resources, for which the techniques developed
proximately 10% with instantaneous penetrations of 25% expe- to date for DG will provide a solid foundation. For example,
rienced at high wind power output. In the U.K., DNOs are fore- the distribution system planning and operation issues of elec-
casting the connection of a further 10 GW of total DG capacity tric vehicles are already the subject of several research studies
by 2015 [74], almost a quarter of the total new generation ca- [82]–[84], even though they are at a very early stage in their trial
pacity expected by the system operator. This scenario requires deployment in power networks and hence there remain many
DG to be considered in transmission planning and operation. unanswered questions about their characteristics.
The network planning methods developed to date have not ad-
dressed this issue and indeed it is non-trivial. It will require more V. CONCLUSION
integrated transmission and distribution models to properly as- The rapid onset of DG in its various forms and scales is trans-
sess the challenges and opportunities. In light of this, the focus forming the traditional planning and operation of distribution
on more traditional objectives such as losses is relevant but may networks. The range of energy resources is matched by the range
have to be adapted to take account of wider system requirements of computational methods and approaches employed in their in-
beyond the distribution network level. tegration. As outlined in this paper, each has their advantages
Ancillary services such as reserve, reactive power and iner- and disadvantages and their appropriate use is dependent on the
tial response are becoming the focus of much attention in re- particular case.
cent times. These services which are not “ancillary” but rather Many objectives have been pursued in the optimization of
vital will in the future have to come from alternative sources the planning of DG. Some of the most common include energy
other than the traditional bulk synchronous generation plant. losses, maximization of DG capacity or energy via sizing and
The provision of such services from DG and the impact of DG siting of DG, minimizing curtailment, or minimizing cost, of-
at transmission is the focus of current attention by researchers tentimes the reinforcement cost associated with DG. One of the
[75]–[80]. From this work it appears that there is significant un- drivers behind the appropriate objective is often the DG owner-
tapped capability from DG but also that additional sources of ship model assumed. For example, a privately owned DG may
support may be required at transmission level [81]. care little about losses but would have a strong preference for
Depending on the particular circumstances of a DG devel- the maximization of DG output. Multi-objective programming
opment, such as resource availability, planning consents or de- methods have tackled a number of these objectives simultane-
clared net capacity, it might be possible to have more than one ously, in an attempt to show the range of “compromise” solu-
network integration scheme (i.e., connection point and/or oper- tions that may be possible.
ation strategy) that is economically sound for the DG developer. The methods developed through research are in some cases
The economics of different locations becomes even more rele- now feeding more directly into the methods employed by net-
vant if not only are infrastructure costs involved but also distri- work operators, or at least pointing the way towards the poten-
bution connection charges. Indeed, DNOs could tune the latter tial benefits of such approaches. Numerous demonstration and
to steer DG projects towards specific areas where the technical field trial projects led in many cases by network operators are
and economic impacts on the system are less onerous or even underway across the world, in order to identify the realizable
beneficial. Alternatively, bilateral commercial arrangements be- benefits highlights in the many works cited here. The status of
tween DNOs and DG developers could also provide win-win DG is itself changing, with micro-generation, in particular pho-
situations. However, for DNOs to determine appropriate loca- tovoltaic cells and larger scale DG becoming more prevalent.
tional signals or commercial arrangements they need to inves- As outlined here, this presents challenges in terms of network
tigate how capable their networks are for integrating renewable modeling, e.g., low voltage networks and wider system opera-
or conventional DG. Such initiatives are particularly relevant tion. Such developments point the way for fruitful areas of fur-
where there is private ownership of DG. If the DNO does not ther research and highlight the dynamic situation and challenges
own the DG and hence can only specify if given DG capaci- all stakeholders in power systems are tackling.
ties are permissible, it is not in a position to specify a preferred
overall allocation to maximize capacity, without entering into REFERENCES
commercial arrangements such as those described above. If a [1] C. Wang and M. H. Nehrir, “Analytical approaches for optimal place-
global benefit is identified by an optimization method, this is ment of distributed generation sources in power systems,” IEEE Trans.
often distinct from the regulatory and commercial framework Power Syst., vol. 19, no. 4, pp. 2068–2076, Nov. 2004.
[2] N. Acharya, P. Mahat, and N. Mithulananthan, “An analytical approach
in place and it may not be a trivial matter to translate a calcu- for DG allocation in primary distribution network,” Int. J. Elect. Power
lated potential benefit for the network into a delivered one. Energy Syst., vol. 28, no. 10, pp. 669–678, Dec. 2006.
1500 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO. 2, MAY 2013

[3] L. F. Ochoa, A. Padilha-Feltrin, and G. P. Harrison, “Evaluating dis- [28] K.-H. Kim, Y.-J. Lee, S.-B. Rhee, S.-K. Lee, and S.-K. You, “Dispersed
tributed generation impacts with a multiobjective index,” IEEE Trans. generator placement using fuzzy-GA in distribution systems,” in Proc.
Power Del., vol. 21, no. 3, pp. 1452–1458, Jul. 2006. 2002 IEEE/PES Summer Meeting, vol. 3, pp. 1148–1153.
[4] L. F. Ochoa, A. Padilha-Feltrin, and G. P. Harrison, “Evaluating [29] W. El-Khattam, K. Bhattacharya, Y. G. Hegazy, and M. M. A. Salama,
distributed time-varying generation through a multiobjective index,” “Optimal investment planning for distributed generation in a compet-
IEEE Trans. Power Del., vol. 23, no. 2, pp. 1132–1138, Apr. 2008. itive electricity market,” IEEE Trans. Power Syst., vol. 19, no. 3, pp.
[5] P. Chiradeja and R. Ramakumar, “An approach to quantify the tech- 1674–1684, Aug. 2004.
nical benefits of distributed generation,” IEEE Trans. Energy Convers., [30] G. Celli, E. Ghiani, S. Mocci, and F. Pilo, “A multiobjective evolu-
vol. 19, no. 4, pp. 764–773, Dec. 2004. tionary algorithm for the sizing and siting of distributed generation,”
[6] Open Distribution System Simulator—OpenDSS. [Online]. Available: IEEE Trans. Power Syst., vol. 20, no. 2, pp. 750–757, May 2005.
http://sourceforge.net/apps/mediawiki/electricdss. [31] G. Celli, E. Ghiani, M. Loddo, F. Pilo, and S. Pani, “Optimal location
[7] Q. Zhou and J. W. Bialek, “Generation curtailment to manage voltage of biogas and biomass generation plants,” in Proc. 2008 Universities
constraints in distribution networks,” IET Proc. Gen., Transm., Dis- Power Engineering Conf. (UPEC), pp. 1–6.
trib., vol. 1, no. 3, pp. 492–498, 2007. [32] E. Zitzler, M. Laumanns, and L. Thiele, Computer Engineering and
[8] C. Dent, L. F. Ochoa, G. P. Harrison, and J. W. Bialek, “Efficient secure Communication Networks Lab (TIK), Swiss Federal Institute of Tech-
AC OPF for network generation capacity assessment,” IEEE Trans. nology (ETH), SPEA2: Improving the strength Pareto evolutionary al-
Power Syst., vol. 25, no. 1, pp. 575–583, Feb. 2010. gorithm, Zurich, Switzerland, 2001, Tech. Rep. 103.
[9] S. N. Liew and G. Strbac, “Maximising penetration of wind generation [33] E. Haesen, J. Driesen, and R. Belmans, “Robust planning method-
in existing distribution networks,” IEE Proc. Gen., Transm., Distrib., ology for integration of stochastic generators in distribution grids,” IET
vol. 149, no. 3, pp. 256–262, May 2002. Renew. Power Gen., vol. 1, no. 1, pp. 25–32, Mar. 2007.
[10] A. Keane and M. O’Malley, “Optimal allocation of embedded gener- [34] T. Sutthibun and P. Bhasaputra, “Multi-objective optimal distributed
ation on distribution networks,” IEEE Trans. Power Syst., vol. 20, no. generation placement using simulated annealing,” in Proc. 2010 Int.
3, pp. 1640–1646, Aug. 2005. Conf. Electrical Engineering/Electronics Computer Telecommunica-
[11] A. Keane and M. O’Malley, “Optimal utilization of distribution net- tions and Information Technology (ECTI-CON), pp. 810–813.
works for energy harvesting,” IEEE Trans. Power Syst., vol. 22, no. 1, [35] M. Sedighi, A. Igderi, and A. Parastar, “Sitting and sizing of distributed
pp. 467–475, Feb. 2007. generation in distribution network to improve of several parameters by
[12] A. Keane, Q. Zhou, J. W. Bialek, and M. O’Malley, “Planning and PSO algorithm,” in Proc. 2010 IPEC, pp. 1083–1087.
operating non-firm distributed generation,” IET Renew. Power Gen., [36] F. S. Abu-Mouti and M. E. El-Hawary, “Optimal distributed generation
vol. 3, no. 4, pp. 455–464, Dec. 2009. allocation and sizing in distribution systems via artificial bee colony
[13] H. W. Dommel and W. F. Tinney, “Optimal power flow solutions,” algorithm,” IEEE Trans. Power Del., vol. 26, no. 4, pp. 2090–2101,
IEEE Trans. Power App. Syst., vol. PAS-87, no. 10, pp. 1866–1876, Oct. 2011.
Oct. 1968. [37] M. F. Sohi, M. Shirdel, and A. Javidaneh, “Applying BCO algorithm
[14] N. S. Rau and Y.-H. Wan, “Optimum location of resources in to solve the optimal DG placement and sizing problem,” in Proc.
distributed planning,” IEEE Trans. Power Syst., vol. 9, no. 4, pp. 2011 5th Int. Power Engineering and Optimization Conf. (PEOCO),
2014–2020, Nov. 1994. pp. 71–76.
[15] G. P. Harrison and A. R. Wallace, “Optimal power flow evaluation of [38] K. Deb, Multi-Objective Optimization Using Evolutionary Algo-
distribution network capacity for the connection of distributed genera- rithms. New York: Wiley, 2001.
tion,” IEE Proc. Gen., Transm., Distrib., vol. 152, no. 1, pp. 115–122, [39] G. P. Harrison, A. Piccolo, P. Siano, and A. R. Wallace, “Exploring
Jan. 2005. the trade-offs between incentives for distributed generation developers
[16] P. N. Vovos, G. P. Harrison, A. R. Wallace, and J. W. Bialek, “Optimal and DNOs,” IEEE Trans. Power Syst., vol. 22, no. 2, pp. 821–828, May
power flow as a tool for fault level constrained network capacity anal- 2007.
ysis,” IEEE Trans. Power Syst., vol. 20, no. 2, pp. 734–741, May 2005. [40] A. Alarcon-Rodriguez, G. W. Ault, and S. J. Galloway, “Multi-objec-
[17] L. F. Ochoa, C. Dent, and G. P. Harrison, “Distribution network ca- tive planning of distributed energy resources: A review of the state-of-
pacity assessment: Variable DG and active networks,” IEEE Trans. the-art,” Renew. Sustain. Energy Rev., vol. 14, no. 5, pp. 1353–1366,
Power Syst., vol. 25, no. 1, pp. 87–95, Feb. 2010. Jun. 2010.
[18] P. N. Vovos and J. W. Bialek, “Direct incorporation of fault level con- [41] K. Deb, A. Pratap, S. Agarwal, and T. Meyarivan, “A fast and elitist
straints in optimal power flow as a tool for network capacity analysis,” multiobjective genetic algorithm: NSGA-II,” IIEEE Trans. Evol. Com-
IEEE Trans. Power Syst., vol. 20, no. 4, pp. 2125–2134, Nov. 2005. putat., vol. 6, no. 2, pp. 182–197, Apr. 2002.
[19] C. Dent, L. F. Ochoa, and G. P. Harrison, “Network distributed gener- [42] E. Haesen, J. Driesen, and R. Belmans, “Long-term planning for small-
ation capacity analysis using OPF with voltage step constraints,” IEEE scale energy storage units,” in Proc. 2007 Int. Conf. Electricity Distri-
Trans. Power Syst., vol. 25, no. 1, pp. 296–304, Feb. 2010. bution (CIRED).
[20] L. F. Ochoa and G. P. Harrison, “Minimizing energy losses: Optimal [43] A. Alarcon-Rodriguez, E. Haesen, G. W. Ault, J. Driesen, and R. Bel-
accommodation and smart operation of renewable distributed genera- mans, “Multi-objective planning framework for stochastic and control-
tion,” IEEE Trans. Power Syst., vol. 26, no. 1, pp. 198–205, Feb. 2011. lable distributed energy resources,” IET Renew. Power Gen., vol. 3, no.
[21] Y. M. Atwa, E. F. El-Saadany, M. M. A. Salama, and R. Seethapathy, 2, pp. 227–238, Jun. 2009.
“Optimal renewable resources mix for distribution system energy loss [44] A. Alarcon-Rodriguez and G. Ault, “Multi-objective planning of dis-
minimization,” IEEE Trans. Power Syst., vol. 25, no. 1, pp. 360–370, tributed energy resources with probabilistic constraints,” in Proc. 2010
Feb. 2010. IEEE/PES General Meeting, p. 7.
[22] G. Celli, S. Mocci, F. Pilo, and G. G. Soma, “A multi-objective ap- [45] X. Pelet, D. Favrat, and G. Leyland, “Multi-objective optimization of
proach for the optimal distributed generation allocation with environ- integrated energy systems for remote communities considering eco-
mental constraints,” in Proc. 2008 Conf. Probabilistic Methods Applied nomics and CO2 emissions,” Int. J. Therm. Sci., vol. 44, no. 12, pp.
to Power Systems (PMAPS). 1180–1189, Dec. 2005.
[23] A. Silvestri, A. Berizzi, and S. Buonanno, “Distributed generation plan- [46] L. F. Ochoa, A. Padilha-Feltrin, and G. P. Harrison, “Time-series based
ning using genetic algorithms,” in Proc. 1999 IEEE/PES PowerTech, maximization of distributed wind power generation integration,” IEEE
p. 257. Trans. Energy Convers., vol. 23, no. 3, pp. 968–974, Sep. 2008.
[24] G. Celli and F. Pilo, “Optimal distributed generation allocation in MV [47] M.-R. Haghifam, H. Falaghi, and O. P. Malik, “Risk-based distributed
distribution networks,” in Proc. 2001 Power Industry Computer Appli- generation placement,” IET Proc. Gen., Transm., Distrib., vol. 2, no.
cations (PICA), pp. 81–86. 2, pp. 252–260, Mar. 2008.
[25] G. Carpinelli, G. Celli, F. Pilo, and A. Russo, “Distributed generation [48] W. S. Andrade, C. L. T. Borges, and D. M. Falcao, “Modeling relia-
siting and sizing under uncertainty,” in Proc. 2001 IEEE/PES Pow- bility aspects of distributed generation connected to distribution sys-
erTech, p. 7. tems,” in Proc. 2006 IEEE/PES General Meeting, p. 6.
[26] G. Carpinelli, G. Celli, S. Mocci, F. Pilo, and A. Russo, “Optimisation [49] A. P. Leite, C. L. T. Borges, and D. M. Falcao, “Probabilistic wind
of embedded generation sizing and siting by using a double trade-off farms generation model for reliability studies applied to Brazilian
method,” IEE Proc. Gen., Transm., Distrib., vol. 152, no. 4, pp. sites,” IEEE Trans. Power Syst., vol. 21, no. 4, pp. 1493–1501, Nov.
503–513, Jul. 2005. 2006.
[27] C. L. T. Borges and D. M. Falcao, “Optimal distributed generation al- [50] C. L. T. Borges and R. J. Pinto, “Small hydro power plants energy avail-
location for reliability, losses, and voltage improvement,” Int. J. Elect. ability modeling for generation reliability evaluation,” IEEE Trans.
Power Energy Syst., vol. 28, no. 6, pp. 413–420, Jul. 2006. Power Syst., vol. 23, no. 3, pp. 1125–1135, Aug. 2008.
KEANE et al.: STATE-OF-THE-ART TECHNIQUES AND CHALLENGES AHEAD FOR DISTRIBUTED GENERATION PLANNING AND OPTIMIZATION 1501

[51] P. Jeongje, L. Wu, C. Jaeseok, A. A. El-Keib, M. Shahidehpour, and [75] S. Grijalva and A. M. J. Visnesky, “Assessment of distributed genera-
R. Billinton, “A probabilistic reliability evaluation of a power system tion programs based on transmission security benefits,” in Proc. 2005
including solar/photovoltaic cell generator,” in Proc. 2009 IEEE/PES IEEE/PES General Meeting, vol. 2, pp. 1441–1446.
General Meeting, p. 6. [76] J. H. Zhao, J. Foster, Z. Y. Dong, and K. P. Wong, “Flexible trans-
[52] L. F. Rocha and C. L. T. Borges, “Probabilistic generation and interrup- mission network planning considering distributed generation impacts,”
tion costs and other economic aspects related to distributed generation IEEE Trans. Power Syst., vol. 26, no. 3, pp. 1434–1443, Aug. 2011.
integration,” in Proc. 2010 IEEE/PES General Meeting, p. 6. [77] P. Cuffe, P. Smith, and A. Keane, “Effect of energy harvesting network
[53] C. L. T. Borges and E. Cantarino, “Microgrids reliability evaluation reactive support on transmission system voltage performance,” in Proc.
with renewable distributed generation and storage systems,” in Proc. 2011 Int. Conf. Exhib. Electricity Distribution (CIRED), pp. 1–4.
2011 IFAC World Congr.. [78] D. E. Newman, B. A. Carreras, M. Kirchner, and I. Dobson, “The im-
[54] V. F. Martins and C. L. T. Borges, “Active distribution network inte- pact of distributed generation on power transmission grid dynamics,”
grated planning incorporating distributed generation and load response in Proc. 2011 Hawaii Int. Conf. System Sciences (HICSS), pp. 1–8.
uncertainties,” IEEE Trans. Power Syst., vol. 26, no. 4, pp. 2164–2172, [79] A. Keane, L. F. Ochoa, E. Vittal, C. Dent, and G. P. Harrison, “En-
Nov. 2011. hanced utilization of voltage control resources with distributed genera-
[55] R. C. Green, L. Wang, M. Alam, C. Singh, and S. S. S. R. Depuru, tion,” IEEE Trans. Power Syst., vol. 26, no. 1, pp. 252–260, Feb. 2011.
“An examination of artificial immune system optimization in intelli- [80] L. F. Ochoa, A. Keane, and G. P. Harrison, “Minimizing the reac-
gent state space pruning for LOLP estimation,” in Proc. 2011 North tive support for distributed generation: Enhanced passive operation and
American Power Symp. (NAPS), p. 7. smart distribution networks,” IEEE Trans. Power Syst., vol. 26, no. 4,
[56] G. W. Ault and J. R. McDonald, “Planning for distributed generation pp. 2134–2142, Nov. 2011.
within distribution networks in restructured electricity markets,” IEEE [81] M. Reza, J. G. Slootweg, P. H. Schavemaker, W. L. Kling, and L. van
Power Eng. Rev., vol. 20, no. 2, pp. 52–54, Feb. 2000. der Sluis, “Investigating impacts of distributed generation on transmis-
[57] V. H. Mendez, J. Rivier, and T. Gomez, “Assessment of energy dis- sion system stability,” in Proc. 2003 IEEE/PES PowerTech, vol. 2, p.
tribution losses for increasing penetration of distributed generation,” 7.
IEEE Trans. Power Syst., vol. 21, no. 2, pp. 533–540, May 2006. [82] K. Clement-Nyns, E. Haesen, and J. Driesen, “The impact of charging
[58] R. A. Walling, R. Saint, R. C. Dugan, J. Burke, and L. A. Kojovic, plug-in hybrid electric vehicles on a residential distribution grid,” IEEE
“Summary of distributed resources impact on power delivery systems,” Trans. Power Syst., vol. 25, no. 1, pp. 371–380, Feb. 2010.
IEEE Trans. Power Del., vol. 23, no. 3, pp. 1636–1644, Jul. 2008. [83] J. A. P. Lopes, F. J. Soares, and P. M. R. Almeida, “Integration of
[59] Recommendations for the Connection of Embedded Generating Plant electric vehicles in the electric power system,” Proc. IEEE, vol. 99,
to Public Distribution Systems Above 20 kV or With Outputs Over no. 1, pp. 168–183, Jan. 2011.
5MW, Electricity Association, 2002, Engineering Recommendation [84] L. P. Fernandez, T. G. S. Roman, R. Cossent, C. M. Domingo, and P.
G75/1. Frias, “Assessment of the impact of plug-in electric vehicles on distri-
[60] IEEE Standard for Interconnecting Distributed Resources With Elec- bution networks,” IEEE Trans. Power Syst., vol. 26, no. 1, pp. 206–213,
tric Power Systems, IEEE Std. 1547-2003, 2003. Feb. 2011.
[61] G. W. Ault, J. R. McDonald, and G. M. Burt, “Strategic analysis frame-
work for evaluating distributed generation and utility strategies,” IEE
Proc. Gen., Transm., Distrib., vol. 150, no. 4, pp. 475–481, Jul. 2003.
[62] R. A. F. Currie, G. W. Ault, C. E. T. Foote, and J. R. McDonald, “Active
Andrew Keane (M’07) received the B.E. and Ph.D. degrees in electrical en-
power-flow management utilising operating margins for the increased
connection of distributed generation,” IET Proc. Gen., Transm., Dis- gineering from University College Dublin, Dublin, Ireland, in 2003 and 2007,
trib., vol. 1, no. 1, pp. 197–202, Jan. 2007. respectively.
[63] R. A. F. Currie, G. W. Ault, R. W. Fordyce, D. F. MacLeman, M. He is a Lecturer with the School of Electrical, Electronic and Communica-
Smith, and J. R. McDonald, “Actively managing wind farm power tions Engineering, University College Dublin with research interests in power
output,” IEEE Trans. Power Syst., vol. 23, no. 3, pp. 1523–1524, Aug. systems planning and operation, distributed energy resources, and distribution
2008. networks.
[64] R. MacDonald and G. W. Ault, “An optimisation model to integrate
active network management into the distribution network investment
planning task,” in Proc. 2011 Int. Conf. Electricity Distribution
(CIRED). Luis F. Ochoa (S’01–M’07–SM’12) received the B.Eng. degree from UNI,
[65] E. Valigi and E. Di Marino, “Networks optimization with advanced Lima, Peru, in 2000 and the M.Sc. and Ph.D. degrees from UNESP, Ilha Solteira,
meter infrastructure and smart meters,” in Proc. 2009 Int. Conf. Elec- Brazil, in 2003 and 2006, respectively.
tricity Distribution (CIRED), p. 4. He is a Lecturer in Smart Distribution Networks at The University of Man-
[66] A. Abart, A. Lugmair, and A. Schenk, “Smart metering features for chester, Manchester, U.K. His current research interests include network inte-
managing low voltage distribution grids,” in Proc. 2009 Int. Conf. Elec- gration of distributed energy resources and future low-carbon distribution net-
tricity Distribution (CIRED), p. 4. works.
[67] E. M. Davidson, S. McArthur, C. Yuen, and M. Larsson, “AuRA-NMS:
Towards the delivery of smarter distribution networks through the ap-
plication of multi-agent systems technology,” in Proc. 2008 IEEE/PES
General Meeting, p. 6.
Carmen L. T. Borges (SM’06) received the B.Sc. degree from the Rio de
[68] T. Boehme, G. P. Harrison, and A. R. Wallace, “Assessment of distribu-
Janeiro State University (UERJ), Rio de Janeiro, Brazil, in 1984 and the M.Sc.
tion network limits for non-firm connection of renewable generation,”
and D.Sc. degrees from the Federal University of Rio de Janeiro (COPPE/UFRJ)
IET Proc. Renew. Power Gen., vol. 4, no. 1, pp. 64–74, Jan. 2010.
in 1991 and 1998, respectively.
[69] G. Celli, E. Ghiani, S. Mocci, F. Pilo, G. Pisano, and G. G. Soma,
She has worked as a Researcher at the Federal University of Rio de Janeiro
“From passive to active distribution networks: Methods and model for
since 1988 and as a Professor of Electrical Engineering since 1996. Her general
planning network transition and development,” in Proc. 2008 CIGRE
research interest is in the area of computer methods for power system analysis
General Session.
and optimization, reliability and high performance computing.
[70] G. Celli, F. Pilo, G. Pisano, and G. G. Soma, “Optimal planning of ac-
Dr. Borges is the PES Chapter Chair of the IEEE Rio de Janeiro Session from
tive networks,” in Proc. 2008 16th Power Systems Computation Conf.
2012 to 2014.
(PSCC’08), p. 8.
[71] American Recovery and Reinvestment Act of 2009, Public Law 111-5,
2009.
[72] Low Carbon Networks Fund Governance Document v.1, Office of Gas
and Electricity Markets (Ofgem), 2010. Graham W. Ault (M’98) received the B.Eng. degree in electrical and mechan-
[73] Low Carbon Networks Fund—Project List, Office of Gas and Elec- ical engineering and the Ph.D. degree in electrical power systems from the Uni-
tricity Markets (Ofgem). [Online]. Available: http://www.ofgem. versity of Strathclyde, Glasgow, U.K., in 1993 and 2000, respectively.
gov.uk/Networks/ElecDist/lcnf/Pages/lcnf.aspx. Since 1996, he has been researching power system planning and operations
[74] Electricity Distribution Price Control Review—Final Proposals—In- issues relating to distributed energy resources in distribution systems. He is cur-
centives and Obligations, Office of Gas and Electricity Markets rently a Professor in the Institute for Energy and Environment at the University
(Ofgem), 2009. of Strathclyde.
1502 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO. 2, MAY 2013

Arturo D. Alarcon-Rodriguez received the Bachelors degree in electro- Chris Dent (M’08) received the B.A. degree in mathematics from Cambridge
mechanical engineering from the Universidad Privada Boliviana, Cochabamba, University, Cambridge, U.K., in 1997, the Ph.D. degree in theoretical physics
Bolivia, in 2001 and the M.Sc. degree in electrical power engineering with from Loughborough University, Loughborough, U.K., in 2001, and the M.Sc.
business and the Ph.D. degree in electrical engineering from the University of degree in operational research from the University of Edinburgh, Edinburgh,
Strathclyde, Glasgow, U.K., in 2005 and 2009, respectively. U.K., in 2006.
From 2008 to 2010, he was a Research Assistant at the Institute of Energy Since 2011 he has been Lecturer in Energy Systems Modelling in the School
and Environment at the University of Strathclyde. His research interests include of Engineering and Computing Sciences, Durham University, Durham, U.K.,
distributed energy resources, power systems planning, and multi-objective opti- having been a Research Fellow at Durham from 2009–2011. From 2007–2009,
mization methods. Currently he works at the Inter-American Development Bank he was with the University of Edinburgh. His research interests lie in power
as an Energy Specialist. system optimization, risk modeling, economics, and renewables integration.

Robert A. F. Currie received the B.Eng. degree in mechanical engineering and Gareth P. Harrison (M’02) is Bert Whittington Chair of Electrical Power En-
business management, the M.Sc. degree in energy systems and the environment, gineering at the University of Edinburgh, Edinburgh, U.K. His current research
and the Ph.D. degree in electronic and electrical engineering from the University interests include network integration of renewable generation and analysis of
of Strathclyde, Glasgow, U.K., in 2001, 2002, and 2010, respectively. the impact of climate change on the electricity industry.
Since 2003, he has been researching, developing, and deploying active net- Prof. Harrison is a Chartered Engineer and member of the Institution of En-
work management solutions to enable distributed generation. He co-founded gineering and Technology.
Smarter Grid Solutions Ltd. in 2008 where he is currently Technical Director.

Fabrizio Pilo (M’98) was born in Sassari, Italy, in 1966. He received the degree
in electrical engineering at the University of Cagliari, Cagliari, Italy, in 1992 and
the Ph.D. degree at the University of Pisa, Pisa, Italy, in 1998.
Currently he is an Associate Professor of Power Systems and Director of
DIEE. His current research interests include electrical power systems, network
planning, and optimization.
Dr. Pilo is an AEIT member, expert of CIGRE SC C6, and Chairman of the
CIRED S5.