Ahmad 2018
Ahmad 2018
Ahmad 2018
1937-9234 © 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
[6]–[8]. Cherukuri and Corts [6] have proposed a relaxation of discrete time instants that continuously sets the generator’s set
ED (RED) problem using a distributed Laplacian-based first- points. This is a suboptimal solution because AGC depends on
order dynamics, which provides exponential convergence. The ED, which is slower, while load keeps on changing on a faster
solution approach in [6] uses a non-smooth penalty function time scale. The above-mentioned limitations demand for new
based approach to map the RED problem as an unconstrained solution approaches to the ED problem, which can respond to
optimization problem and performs a distributed allocation of fast fluctuations in the generation as well as load.
loads to the set of DGs. The solution approach in [7] offers a To address this issue, the conventional primal-dual algorithm
distributed consensus control for optimized management of dis- based ED is extended in this paper. Specifically, to achieve
tributed generation resources for an islanded mode of the micro- improved dynamic response, an augmented Lagrangian
grid operation. The above-mentioned solution approaches suffer based dynamic control is proposed. The augmentation to the
from the limited dynamic performance of economic dispatch. Lagrangian function allows us to implement a proportional–
Microgrid employs a hierarchical control strategy for its integral–derivative (PID) controller. In addition, for distributed
operation. At the bottom of the hierarchy is the primary control implementation, a multiagent-based system architecture
control that maintains the operating voltage and frequency. The is proposed. Traditionally, a multiagent-based architecture
secondary control is responsible for active and reactive power converges to the average consensus [15], which in case of
controls, which is also termed as automatic generation control distributed generation can provide equal load sharing or equiv-
(AGC). To minimize the generation cost while meeting the alently consensus among the DGs. This is a useful attribute for
load demands, we need to operate different DGs at the optimal controlling DGs of equal rating. On the other hand, real-time
power generation point. For that purpose, tertiary control is ED can always require unequal load sharing among the DGs.
used to implement economic dispatch. To achieve any desired level of tradeoff between ED and
Conventionally, the ED problem is performed on a slower generation consensus among DGs, a relaxation of consensus
time scale. Although the conventional approaches to solve ED constraint is introduced. The proposed solution approach is
problem have been efficacious for traditional power systems, equally useful for grid connected as well as islanded modes of
they fall short for a microgrid system with frequent transients operation. However, it is more effective for an islanded mode
and larger power fluctuations. This can be attributed to the fact of operation, in which case transients can be more pronounced.
that the load demand uncertainties as well as the number of Normally, ED problem and dynamic response problem are
renewable resources are on the rise. As a result, the generation addressed separately in the literature. One key contribution of
system is exposed to large power fluctuations [9], resulting in this paper is to embed dynamic response in ED optimization. In
frequent transients. Cost-based droop schemes for economic addition, some other contributions of this paper are as follows:
dispatch in islanded microgrids are presented in [10]. A robust 1) A fully distributed augmented Lagrangian-based PID con-
two-stage approximate optimization approach is presented to trol algorithm that can be applied to an islanded microgrid
schedule the energy generation under uncertainties to minimize to improve the dynamic performance during transients;
long-term average operating cost in [11]. The first stage of 2) Optimal power sharing of the generators;
optimization determines hourly unit commitment of the con- 3) Problem formulation for an economic-emission dispatch
ventional generators (CGs) via a day-ahead scheduling, and the with PV-integrated transient control with distributed
second stage performs economic dispatch of the CGs, energy loads, such as electric vehicles (EVs).
storage systems, and energy trading via a real-time scheduling. This paper is organized as follows. In Section II-A, a mi-
Though the solution handles the uncertainties due to renewable crogrid system architecture with multiple generation systems is
generation and load demands, however, it offers an approximate outlined. Details of the distributed control with the multiagent
optimization solution for the long-term operational cost. Peak- architecture is presented in Section II-B. The ED based on the
aware online economic dispatching for microgrids is studied in conventional and proposed PID control along with emission re-
[12] considering competitive online algorithms that do not rely ductions and integration of renewable energy resources is given
on prediction of the future input. System dynamics or transients, in Section III. Performance evaluation results are provided in
however, have not been considered in this paper. The focus Section IV and this paper is concluded in Section V.
was only on cost reduction. A two-stage control framework
and a hybrid control method are proposed for active power II. SYSTEM MODEL
real-time dispatch in islanded ac microgrids in [13]. However,
the proposed solution suffers from slow dynamic response. A. Microgrid System Architecture
The primal-dual algorithm based distributed ED solution also Microgrids are classified based on their architectures, con-
suffers from poor dynamic performance due to the presence of trol, and communication interfaces as residential, industrial, or
inherent integral controller as will be explained in Section III-A. military microgrids. Our proposed microgrid system architec-
In addition, the time gap between slower ED and faster AGC ture is shown in Fig. 1 consisting of a mix of DGs of differ-
leads to reduced economical efficiency of the microgrid. Few ent types. In this model, renewable and nonrenewable energy
studies have proposed to bridge this gap between ED and sources are combined together in the form of a hybrid microgrid.
AGC. Integrating AGC and ED for real-time optimization is DC–DC converters, dc–ac inverters, power management sys-
proposed in [14] and ED in the feedback loop is activated at tem, and controllable loads are some other system components
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
AHMAD et al.: DYNAMIC ECONOMIC DISPATCH AND TRANSIENT CONTROL OF DISTRIBUTED GENERATORS IN A MICROGRID 3
dynamic performance, an augmented Lagrangian based PID equations can be rewritten as follows:
controller for ED will be proposed. +
t
u i = Φ Mi −
t
kλ Ld − zi (τ ) dτ + uρ − uσ
A. Conventional ED 0 i
The ED optimization problem with a quadratic cost function t
Ci is defined as uρ = kρ i (zm in − zi (τ ))+ dτ
0
Ci (zi ) = αi zi2 + βi zi + γi ∀i. (1) t
i i uσ = − kσ i (zi (τ ) − zm ax )+ dτ. (5)
0
In (1), zi ∈ z, z ∈ RN , represents the power delivered form The auxiliary variables consisting of ui , uρ , and uσ in (5) are
generator i, whereas αi , βi , and γi are cost coefficients of the in fact part of the control law, which effectively implements an
ith generator. The ED optimization problem then becomes integral control to adjust power generation from each DG.
minimize ωi Ci (zi )
B. Proposed PID Based ED
i
Due to the inherent integral control action, the dynamic
s.t. zi ≥ Ld , Mz + δ ≥ 0, zm in ≤ zi ≤ zm ax . (2)
performance of the optimized power generation may not be
i
satisfactory. To improve the dynamic performance during
In (2), M ∈ RN ×N . ωi in the objective function is the weighting power transients due to any variations in load demand, a
coefficient, whereas Ld denotes the load demand. The second dynamic performance controller based solution is developed.
inequality constraint is generator’s power limits between zm in This is achieved by modifying the system dynamics given in
and zm ax . The third equality constraint in (2) is termed as (5), to incorporate additional terms. For that purpose, we extend
the consensus constraint. When Mz + δ = 0, all generators the Lagrangian in (3) to construct an augmented Lagrangian
reach at a consensus. By allowing δ ≥ 0 and correspondingly function La given by
Mz + δ ≥ 0, consensus constraint is relaxed, which in turn 2
provides the flexibility to reduce the overall generation cost. To kp
La (z, λ, Φ, ρ, σ, z̃) = ωi Ci (zi ) + Ld − zi
map the constrained optimization problem in (2) to an equiva- i i
2 i
lent unconstrained problem, we use the Lagrange function. The
Lagrangian for the optimization problem in (2) is defined as
+ ki λ Ld − zi
i
L(z, λ, Φ, ρ, σ) = ωi Ci (zi ) + λ Ld − zi
i i + Φt [Mz + δ] + ρ(zm in − zi )
i
+ Φt [Mz + δ] + ρ(zm in − zi )
kd
i
+ σ(zi − zm ax ) + (zi − z̃i )2 .
2
+ σ(zi − zm ax ). (3) i i
i
(6)
In (3), λ, ρ, σ, and Φ are the Lagrange multipliers (or dual vari- In (6) kp , ki , and kd are, respectively, proportional, integral,
ables) associated with the load demand, consensus constraint, and derivative gains of the controller and z̃i is an auxiliary state
and individual generator’s minimum and maximum power lim- variable. Using the augmented Lagrangian in (6), the updated
its, respectively. Using (3), next, we construct the primal-dual primal-dual dynamics and associated control law ui becomes
dynamics, for the optimization problem in (2), given as follows: żi = kz i (ωi Ci (zi ) + ui ) , z̃˙i = k̃d (zi − z̃i )
żi = kz i (ωi Ci (zi )
+ Φ Mi − λ − ρ + σ) t +
+ λ̇ = kλ Ld − zi , φ̇i = kφ i {[Mz]i + δ}+
λ̇ = kλ Ld − zi , φ̇i = kφ i {[Mz]i + δ}+ i
i ρ̇ = kρ i {zm in − zi }+ , σ̇ = kσ i {zi − zm ax }+
ρ̇ = kρ i {zm in − zi }+ σ̇ = kσ i {zi − zm ax }+ . (4)
ui = −ki [λ + ρ − σ − Φt Mi ]
In (4), Mi represents the ith column of M, [Mz]i denotes the ith
+ kd (zi − z̃i ) − kp ψ(λ̇) ∀i. (7)
element of vector Mz, and φi ∈ Φ. The coefficients kz i , kλ and
kφ i , kρ i , kσ i are controller gains, whereas the notation {z}+ in The first expression in (7) represents the power system dynam-
(4) is defined as max{0, z}. Let ui = Φt Mi − (λ + uρ − uσ ) ics for ED and the associated controller ui . The control law ui
is the control action. Using the expression for λ, ρ, and σ from implements the PID control, where ψ(.) is a linear functional
(4) and substituting to the control action ui , the system dynamic mapping. The first term, in ui , is the integral control action. The
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
AHMAD et al.: DYNAMIC ECONOMIC DISPATCH AND TRANSIENT CONTROL OF DISTRIBUTED GENERATORS IN A MICROGRID 5
TABLE II
COST COEFFICIENTS FOR DGS
Fig. 6. Consensus among DGs for optimal power generation based on (a) conventional control and (b) PID control.
The result of the sigmoid function fit is shown in Fig. 3. The TABLE III
PID CONTROLLER GAINS
function fr (ppv ) is nonconvex, however, log transformation
results in a concave function, which is used to maximize the
reliability.
3) Optimization Problem Formulation: There is an under-
lying tradeoff among the conflicting objectives of pollutant
emission reduction, PV power reliability enhancement, and gen-
eration cost reduction. To achieve this tradeoff, we formulate a
weighted multiobjective optimization problem using the weight-
ing coefficients ωi and ωre , where ωre is the weighting coefficient
to provide tradeoff between PV reliability and emissions. The
multiobjective ED problem can now be reformulated as follows: The connectivity among the four DGs is illustrated in Fig. 5.
1
For the given connectivity among the DGs, the corresponding
min ωi Ci (zi ) + Ei (zi ) − ωre log(fr (ppv )) Laplacian matrix is given by
ωre
i ⎡ ⎤
1 −1 0 0
s.t. zi + ppv = Ld + LEV , zi ≥ Ld ⎢ ⎥
⎢ −1 3 −1 −1 ⎥
i i
M =⎢ ⎢ 0 −1 2 −1 ⎥ .
⎥ (10)
Mz + δ ≥ 0, zm in ≤ zi ≤ zm ax . (9) ⎣ ⎦
0 −1 −1 2
In (9), ppv is the power output from PV panels and LEV is a
distributed load due to electric vehicles, which also affects The power generation capacities for the four generators are
system transients. The problem in (9) can be solved using the tabulated in Table I. It is assumed that the generators are con-
Lagrangian duality, as discussed previously. ventional thermal power units with generator’s cost coefficients
given in Table II. The generator’s cost coefficients in Table II are
IV. PERFORMANCE EVALUATION RESULTS derived from [24], whereas pollutant emission cost coefficients
are taken from [21]. We require a total power demand of 4 MW.
Since this paper is implemented at algorithmic level, most The information among the DGs is exchanged using an IEEE
of the results are obtained through numerical simulations of 802.15.4-based communication interface, which provides a data
the proposed microgrid system. For that purpose, MATLAB rate of 250 Kb/s. The step-size scaling coefficients are config-
R2016a is used on Intel Core i5 laptop. Please refer to Fig. 4 ured with constant values as kp i = 2000, k̃p i = 400, kφ i = 20,
that has been used for simulating the PID dynamics given in and kλi = 30.
(7) and its extended version given by (9) with their descriptions
given in Sections III-B and III-C3, respectively.
A. Case1: Performance of the Conventional and Proposed
For the optimized power flow control from the distributed
generation, we consider a network of four DGs [23]. Each PID Controls With Generator Power Constraints
DG comprises of a controller module, an energy source, and We compare the performance of the optimized PID based
a power converter. The proposed PID algorithm is applicable ED with that of the conventional ED under generator power
for both grid connected as well as islanded mode of operation. constraints and the results are shown in Fig. 6(a) and (b). The
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
AHMAD et al.: DYNAMIC ECONOMIC DISPATCH AND TRANSIENT CONTROL OF DISTRIBUTED GENERATORS IN A MICROGRID 7
Fig. 7. Optimal power generation from different DGs based on the conven-
tional control.
Fig. 8. Optimal power generation from different DGs based on the optimized generator for ED. In addition, the dynamic performance of the
PID control. two controllers is also compared. For this scenario, the param-
eter δ is set equal to one-half of the load demand. This setting
restricts the maximum generated power difference between any
step-size scaling coefficients are configured with constant values
pair of generators not more than 2 MW. The performance of
as k̃ρ i = 400, kφ i = 20, kλ = 30, and kρ i = 200.
integral control for ED is shown in Fig. 7. It can be observed
Introducing generator’s power limits (zm in , zm ax ), conven-
from Fig. 7 that system response exhibits poor transient per-
tional control has more pronounced transients compared to
formance. There is high overshoot at the beginning as system
PID control. With proper tuning, PID control outperforms the
tries to adjust the power from different generation units. It takes
conventional control. The consensus constraint in (2) allows us
approximately 0.7 s to reach steady state. At 2 and 4 s, the
to achieve any desired tradeoff between cost optimization and
system experiences a step change in load demands to 12 Ld and
uniform loading of multiple DGs. The system also achieves a 3
consensus ensuring that all DGs are generating same power. 4 Ld , respectively, resulting in power transients. For these step
changes in the load demand, we observe poor transient perfor-
The tuned values of the PID gains used for this case are
mance again. The same load transient scenario is presented to
tabulated in Table III.
the optimized PID and the response is shown in Fig. 8. The
tuned values of the PID gains used for this case are tabulated
B. Case 2: Performance of PID Without Generator in Table III. From the results in Fig. 8, we observe a significant
Power Constraints
improvement in the transient performance and system settles
In this case, we first compare the performance of integrator- in less than 0.2 s. In addition, the overshoot is also reduced
based controller with the proposed PID based optimized power significantly, as can be seen from Fig. 8.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
Fig. 11. Effect of δ variations on the power allocation to different DGs. (a) δ = 0.8. (b) δ = 0.4. (c) δ = 0.
Fig. 12. (a) PI controller performance with EV load. (b) Optimized controller performance with EV load.
C. Case 3: Effect of δ on Generation Cost reduction arbitrarily. Rather, we do not get any cost reduction
Next, we analyze the effect of parameter δ on the total beyond a certain threshold value of δ, which corresponds to the
generation cost for different load demands. For a given load optimal generation cost and the consensus constraint becomes
demand, increasing δ leads to generation cost reduction, as can inactive at that point. For a fixed load demand, let Cδ denote
be seen from Fig. 9. The cost reduction is based on the fact that a the total generation cost for a given value of δ, whereas Cm in
DG with lower cost contributes more power in contrast to a DG denotes the minimum possible generation cost for the same load
with higher generation cost. Increasing δ does not provide cost demand, then we define incremental generation cost, attributed
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
AHMAD et al.: DYNAMIC ECONOMIC DISPATCH AND TRANSIENT CONTROL OF DISTRIBUTED GENERATORS IN A MICROGRID 9
to consensus, as ΔC = Cδ − Cm in . For different load demands, From these results, it is obvious that an optimized controller has
the incremental generation cost ΔC as a function of δ is shown superior dynamic performance compared to a PI controller.
in Fig. 10. As the parameter δ increases, the incremental
generation cost reduces and becomes zero beyond the threshold E. Case 5: Effect of Pollutant Emission and PV
value of δ. The threshold value for δ increases with an increase Power Reliability
in the load demand as can be observed from Fig. 10. In
Effects of CO2 emission and PV power reliability are also
addition, an increase in the load demand also results in higher
analyzed using the revised optimization problem in (9) and the
incremental cost, due to the generation from higher cost DGs.
results are shown in Fig. 13(a) and (b). Fig. 13(a) shows the
The relaxed consensus constraint in (2) allows to achieve any
increase in CO2 emissions (ton/hour) as a function of parameter
desired tradeoff between cost optimization and uniform loading
ωre for different loads. In addition, Fig. 13(b) shows the PV
of multiple DGs. Setting a large value of parameter δ results
power reliability as a function of parameter ωre . From these
in cost optimization, whereas making it zero ensures that all
results, it is obvious that demanding higher reliability reduces
DGs are generating same power. Next, we analyze the effect of
PV power contribution to the system.
varying parameter δ on the power allocation to different DGs.
This is shown in Fig. 11 for three different values of parameter
δ. Setting a value of 0.8 to parameter δ, the difference between F. Case 6: Impact of Frequency-Dependent and
the powers generated from the DGs connected to each other Voltage-Dependent Loads
does not exceed 0.8 MW, as can be observed from Fig. 11(a). Different frequency- and voltage-dependent load models are
Similarly, when δ is set to zero, all the DGs are generating same studied in the literature [26], [27]. Among those ZIP, exponen-
power as verified from Fig. 11(c). tial, and linear frequency dependent models are widely used
for load modeling. An exponential load model [28] with both
voltage and frequency dependence has been adopted in this
D. Case 4: Effect of Electric Vehicle Load on the Performance study and its impact on transient performance of the micro-
of PI and Optimized Controller grid is studied. The load model, used to analyze voltage- and
To further explore the transient response of the proposed frequency-dependent load variations, is given as follows:
solution, the measured data from [25] is used to construct a k pv
E w − wo
load model of EV charging station for a Nissan Leaf EV, having Ld (E, w) = Po 1 + kpf (11)
Eo wo
50-kW charging load. The results are shown in Fig. 12(a) and (b).
When the charging load of 5 MW for a fleet of Nissan Leaf EVs is where kpv = 1.7 and kpf = 1 and these are taken from [27] for
applied for up to 2 s, the generators share this load. Performance a residential load. Po is nominal load that is taken as 4 MW, E
of PI control is compared with an optimized controller. As shown is system operating voltage, and Eo is nominal voltage. w is
in Fig. 12(a), a PI controller has greater transients compared to the operating frequency and wo is the nominal frequency. The
an optimized controller as shown in Fig. 12(b). After 2 s, the frequency deviation was taken from −1% to 0.6% and its effect
EV load is gradually reduced, the load demand on generators is shown in Fig. 14(a) for frequency-dependent load. The voltage
due to EVs is also adjusted and followed closely by the DGs. deviation was taken from −2% to 2% and the corresponding
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
result is shown in Fig. 14(b). The accumulated power from DGs [14] Y. A. Abass, A. T. Al-Awami, and T. Jamal, “Integrating automatic genera-
follows closely the required voltage-dependent and frequency- tion control and economic dispatch for microgrid real-time optimization,”
in Proc. IEEE Power Energy Soc. Gen. Meeting, 2016, pp. 1–5.
dependent load power, as observed from these results. [15] N. Cai, N. T. T. Nga, and J. Mitra, “Economic dispatch in microgrids using
multi-agent system,” in Proc. North Amer. Power Symp., 2012, pp. 1–5.
[16] J. M. Guerrero, J. C. Vasquez, J. Matas, L. G. de Vicuña, and M. Castilla,
V. CONCLUSION “Hierarchical control of droop-controlled ac and dc microgrids general
approach toward standardization,” IEEE Trans. Ind. Electron., vol. 58,
The problem of the optimized control for ED in microgrid- no. 1, pp. 158–172, Jan. 2011.
based distributed generation is considered. To overcome the [17] A. Bidram, A. Davoudi, F. L. Lewis, and Z. Qu, “Secondary control
performance limitations of integral-based control, PID-based of microgrids based on distributed cooperative control of multi-agent
systems,” IET Gener. Transmiss. Distrib., vol. 7, no. 8, pp. 822–831,
optimized control is proposed that provides an improved dy- Aug. 2013.
namic performance. In particular, the augmented Lagrangian [18] N. M. Dehkordi, N. Sadati, and M. Hamzeh, “Fully distributed cooperative
approach is used to design the dynamic controller providing secondary frequency and voltage control of islanded microgrids,” IEEE
Trans. Energy Convers., vol. 32, no. 2, pp. 675–685, Jun. 2017.
a tight control of system dynamics. In addition, a consensus [19] A. Bidram and A. Davoudi, “Hierarchical structure of microgrids con-
constraint is introduced to tradeoff between cost optimality and trol system,” IEEE Trans. Smart Grid, vol. 3, no. 4, pp. 1963–1976,
power sharing among DGs. The PID performance is compared Dec. 2012.
[20] G. Tina, S. Gagliano, and S. Raiti, “Hybrid solar/wind power system prob-
with the conventional integral based control for ED in a mi- abilistic modelling for long-term performance assessment,” Sol. Energy,
crogrid. Furthermore, electric vehicle type load is included in vol. 80, no. 5, pp. 578–588, 2006.
system to further study load intermittency. Impact of frequency- [21] A. Jubril, O. Olaniyan, O. Komolafe, and P. Ogunbona, “Economic-
emission dispatch problem: A semi-definite programming approach,”
dependent and voltage-dependent loads are also analyzed. Appl. Energy, vol. 134, no. Suppl. C, pp. 446–455, 2014.
The proposed solution can be extended for event-driven ED, [22] H. Kanchev, F. Colas, V. Lazarov, and B. Francois, “Emission reduction
in the multiagent framework, to further reduce the communi- and economical optimization of an urban microgrid operation including
dispatched PV-based active generators,” IEEE Trans. Sustain. Energy,
cation bandwidth requirements. An optimization formulation vol. 5, no. 4, pp. 1397–1405, Oct. 2014.
can also be investigated for a multimicrogrid case with energy [23] M. Tahir and S. K. Mazumder, “Self-triggered communication enabled
routing to meet energy internet requirements. control of distributed generation in microgrids,” IEEE Trans. Ind. Inform.,
vol. 11, no. 2, pp. 441–449, Apr. 2015.
[24] Y. Xu, W. Zhang, and W. Liu, “Distributed dynamic programming-based
REFERENCES approach for economic dispatch in smart grids,” IEEE Trans. Ind. Inform.,
vol. 11, no. 1, pp. 166–175, Feb. 2015.
[1] F. Katiraei and M. R. Iravani, “Power management strategies for a micro- [25] R. C. Leou, J. H. Teng, and C. L. Su, “Modelling and verifying the load
grid with multiple distributed generation units,” IEEE Trans. Power Syst., behaviour of electric vehicle charging stations based on field measure-
vol. 21, no. 4, pp. 1821–1831, Nov. 2006. ments,” IET Gener. Transmiss. Distrib., vol. 9, no. 11, pp. 1112–1119,
[2] M. Jain, S. Gupta, D. Masand, and G. Agnihotri, “Analysis of a mi- 2015.
crogrid under transient conditions using voltage and frequency con- [26] S. Palm and P. Schegner, “Static and transient load models taking account
troller,” Adv. Power Electron., vol. 2012, 2012. [Online]. Available: voltage and frequency dependence,” in Proc. Power Syst. Comput. Conf.,
https://www.hindawi.com/journals/ape/2012/208231/ Jun. 2016, pp. 1–7.
[3] M. L. Fisher, “The Lagrangian relaxation method for solving integer [27] I. T. F. on Load Representation for Dynamic Performance, “Load repre-
programming problems,” Manage. Sci., vol. 50, no. 12, pp. 1861–1871, sentation for dynamic performance analysis [of power systems],” IEEE
2004. [Online]. Available: http://www.jstor.org/stable/30046157 Trans. Power Syst., vol. 8, no. 2, pp. 472–482, May 1993.
[4] C.-L. Chen and N. Chen, “Direct search method for solving economic [28] H. H. Zeineldin and J. L. Kirtley, “Micro-grid operation of inverter based
dispatch problem considering transmission capacity constraints,” IEEE distributed generation with voltage and frequency dependent loads,” in
Trans. Power Syst., vol. 16, no. 4, pp. 764–769, Nov. 2001. Proc. IEEE Power Energy Soc. Gen. Meeting, Jul. 2009, pp. 1–6.
[5] A. G. Tsikalakis and N. D. Hatziargyriou, “Centralized control for op-
timizing microgrids operation,” in Proc. IEEE Power Energy Soc. Gen.
Meeting, Jul. 2011, pp. 1–8.
[6] A. Cherukuri and J. Corts, “Distributed generator coordination for ini-
tialization and anytime optimization in economic dispatch,” IEEE Trans.
Control Netw. Syst., vol. 2, no. 3, pp. 226–237, Sep. 2015.
[7] Y. Xu and Z. Li, “Distributed optimal resource management based on the
consensus algorithm in a microgrid,” IEEE Trans. Ind. Electron., vol. 62,
no. 4, pp. 2584–2592, Apr. 2015.
[8] W. T. Elsayed and E. F. El-Saadany, “A fully decentralized approach for Jameel Ahmad received the B.Sc. degree in elec-
solving the economic dispatch problem,” IEEE Trans. Power Syst., vol. 30, trical engineering from the University of Engineer-
no. 4, pp. 2179–2189, Jul. 2015. ing and Technology (UET) Peshawar, Peshawar,
[9] V. Loia and A. Vaccaro, “Decentralized economic dispatch in smart grids Pakistan, in 1993, and the M.Sc. degree in electrical
by self-organizing dynamic agents,” IEEE Trans. Syst. Man, Cybern., engineering from the University of Southern Califor-
Syst., vol. 44, no. 4, pp. 397–408, Apr. 2014. nia at Los Angeles, Los Angeles, CA, USA, in 2005.
[10] F. Chen et al., “Cost-based droop schemes for economic dispatch in He is currently working toward the Ph.D. degree in
islanded microgrids,” IEEE Trans. Smart Grid, vol. 8, no. 1, pp. 63–74, electrical engineering at the Department of Electrical
Jan. 2017. Engineering, UET Lahore, Lahore, Pakistan, under
[11] W. Hu, P. Wang, and H. B. Gooi, “Toward optimal energy management of the supervision of Dr. M. Tahir.
microgrids via robust two-stage optimization,” IEEE Trans. Smart Grid, He was with Qualcomm and Broadcom Corpora-
vol. 9, no. 2, pp. 1161–1174, Mar. 2018. tion, San Diego, CA, USA, from June 2007 to June 2010, on 2G and 3G mobile
[12] Y. Zhang, M. H. Hajiesmaili, S. Cai, M. Chen, and Q. Zhu, “Peak-aware communication systems. Since August 2010, he has been an Assistant Pro-
online economic dispatching for microgrids,” IEEE Trans. Smart Grid, fessor with the Department of Electrical Engineering, School of Engineering,
vol. 9, no. 1, pp. 323–335, Jan. 2018. University of Management and Technology Lahore, Lahore, Pakistan. His re-
[13] J. Zhao and P. L, “Decentralized two-stage control for active power real search interests include control and optimization of smart microgrids for energy
time dispatch in islanded ac microgrids,” in Proc. IEEE Conf. Energy internet.
Internet Energy Syst. Integr., Nov. 2017, pp. 1–6. Prof. Ahmad is the reviewer of numerous IEEE journals and conferences.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
AHMAD et al.: DYNAMIC ECONOMIC DISPATCH AND TRANSIENT CONTROL OF DISTRIBUTED GENERATORS IN A MICROGRID 11
Muhammad Tahir (M’02) received the Ph.D. de- Sudip K. Mazumder (S’97–M’01–SM’03–F’16) re-
gree in electrical and computer engineering from the ceived the Ph.D. degree in electrical and computer en-
University of Illinois at Chicago, Chicago, IL, USA, gineering from Virginia Tech, Blacksburg, VA, USA,
in 2008. in 2001.
He is currently an Associate Professor with He has been a Professor with the University of
the Department of Electrical Engineering, Univer- Illinois at Chicago (UIC), Chicago, IL, USA, since
sity of Engineering and Technology Lahore, Lahore, 2001, and has been the President of NextWatt LLC
Pakistan. He was a Visiting Research Scholar with since 2008. He has more than 25 years of professional
the University of Illinois at Chicago in the summer of experience and has held R&D and design positions
2014. He was the principal investigator on the funded in leading industrial organizations and has served as
research projects related to air quality monitoring and a Technical Consultant for several industries. He has
intelligent camera networks. He has more than 40 international publications to authored or coauthored more than 200 refereed papers, delivered more than 85
his credit. His research interests include wireless sensor networks, delay-tolerant keynote/plenary/invited presentations, and received and carried out about 50
networks, distributed resource optimization for wireless networks, and real-time sponsored research projects since joining UIC. He is the author or co-author of
wireless multimedia networks. more than 180 journal and conference papers and 9 book/book chapters, and
Dr. Tahir is the reviewer of numerous IEEE journals and conferences. holds 8 issued and several pending patents.
Dr. Mazumder was the Guest Editor-in-Chief and an Associate Editor for
multiple IEEE Transactions and the first Editor-in-Chief for Advances in Power
Electronics. He has served and is serving in high-profile capacities in leading
IEEE conferences, technical committees, and NSF panels. He was the recipient
of several prestigious awards including the UIC’s Inventor of the Year Award
(2014), the University of Illinois’ University Scholar Award (2013), the IEEE
PELS Transaction Prize Paper Award (2001), the IEEE Future Energy Chal-
lenge Award (2005), the ONR Young Investigator Award (2005), and the NSF
CAREER Award (2003).