IEEE TRANSACTIONS ON SMART GRID, VOL. 10, NO.

1, JANUARY 2019 1005

A Circuit Theory-Based Loss Allocation Method

for Active Distribution Systems
Pankaj Kumar, Student Member, IEEE, Nikhil Gupta , Member, IEEE,
Khaleequr Rehman Niazi , Senior Member, IEEE, and Anil Swarnkar, Member, IEEE

Abstract—The integration of optimally placed distributed gen- k, q Node number

eration (DG) units in a distribution system can contribute K(ij, n(ij, k)) Cross terms allocation factor for branch ij
significantly in loss reduction among distribution feeders. In prac- MT(ij) Mixed terms of current in branch ij
tice, this loss reduction may provide associated benefit to load n(ij, k) Nodes beyond branch ij, for k∈CN(ij)
points (consumers) though they are not contributing for the same.
N Total number of nodes in the system
The loss allocation strategy should be such as to benefit DG own-
ers who have actually contributed toward loss reduction. This NB Set of branches of the distribution
paper presents a new circuit theory-based loss allocation method systems
for active distribution systems. The method first employs branch- NDG Total number of DG
oriented approach to decompose cross-terms of power loss among p, r DG number
contributing load points. Thereafter, the contributing currents Ploss Total real power loss in the system
of DGs are determined using superposition principle to allocate Ploss(ij) Real power loss of branch ij
their remuneration/penalty. The proposed method is applied to ploss(ij, n(ij, k)) Real power loss of branch ij allocated to
33-bus test distribution system. The application results highlight consumer connected at node n(ij, k)
the importance of the proposed method.
ploss(k) Real power loss allocated to consumer
Index Terms—Active distribution systems, circuit theory-based connected at kth node
method, distributed generations, loss allocation, superposition. R(ij) Resistance of branch ij
{x} Real part of complex quantity, x
RDG (ij) Remuneration of DG through the
branch ij
RDG (ij, p) Remuneration of pth DG through the
N OMENCLATURE branch ij
cos (nij, k)) Load power factor at the kth node feeding ST(ij) Square terms of current in branch ij
branch ij Vk Voltage at kth node.
CT(ij) Cross terms of current in branch ij
CT(ij,n(ij, k)) Cross terms of each contributing node
current I(n(ij, k)) in branch ij I. I NTRODUCTION
CN(ij)/CDG(ij) Set of contributing node/ DG current in HE CONCEPT of smart grid has completely revolu-

ij
branch ij
Branch number
T tionized the electricity tariff structure. The electricity
tariff structure should be rationalized in such a fashion that
I(ij) Current of branch ij the associated benefit should reach to all stakeholders’ right
I(n(ij, k)) Load current of consumer at node n(ij, k) from consumer end to generation companies. The cost of
Im{x} Imaginary part of complex quantity, x distributing electrical energy is an important component of
IQ(n(ij, k)) Reactive component of load current at network use tariff. Apart from other network use charges,
node n(ij, k) the evaluation of charges incurred for feeder power losses is
IR(n(ij, k)) Active component of load current at node of utmost importance, but is highly challenging on account
n(ij, k) of non-linear relationship between losses and the power
I c (ij) Current in branch ij with DGs delivered.
I DG (ij) DG current in branch ij With the advancement in electricity market, several
Ik Load current at kth node loss allocation methods have been proposed. The major-
I u (ij) Current in branch ij without DGs ity of these loss allocation procedures include: pro-rata
method [1], [2], quadratic method [3], [4], proportional shar-
ing method [5]–[10], substitution method [11], incremental
Manuscript received March 2, 2017; revised May 10, 2017 and August 21,
2017; accepted September 18, 2017. Date of publication September 27, 2017;
method [12]–[15], direct loss coefficient method [16], circuit
date of current version December 19, 2018. Paper no. TSG-00307-2017. theory-based methods [17]–[20], etc. Pro-rata method is prob-
(Corresponding author: Khaleequr Rehman Niazi.) ably the simplest one as it assigns a definite percentage of
The authors are with the Department of Electrical Engineering, power loss among all users in accordance to their active
Malaviya National Institute of Technology, Jaipur 302017, India
(e-mail: [email protected]; [email protected];
power generation/consumption. However, the method does
[email protected]; [email protected].) not provide true signal for loss allocation as the allocated
Digital Object Identifier 10.1109/TSG.2017.2757059 losses are independent of network topology so the method
1949-3053 c 2017 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.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KANPUR. Downloaded on June 28,2022 at 05:33:30 UTC from IEEE Xplore. Restrictions apply.
 1006 IEEE TRANSACTIONS ON SMART GRID, VOL. 10, NO. 1, JANUARY 2019

has not gained general acceptance [21]. This limitation is participation factors of the participants must be positive and
overcome in quadratic and proportional loss allocation meth- should lie within the range of 0–2. However, these restrictions
ods which consider network topology, but are based upon may be violated in distribution system having disproportionate
heuristic formulations. In substitution methods, losses are cal- sizing of loads or DGs.
culated by determining the difference of system losses while Most of the above mentioned loss allocation methods were
connecting and disconnecting the component. However, the originally suggested for the transmission systems and then
results obtained by this method are found to be inconsistent attempted for distribution systems. However, the nature and
and the method lacks sound economic foundation [16], [22]. behavioral aspects of these two systems are different on
The incremental methods have gained more acceptability to account of typical differences on the basis of R/X ratio of
market environment as they are based upon more appeal- lines, load profiles, network configuration, characteristic of
ing approach of assigning incremental change in power loss loads, reactive power consumed/injected by end users, etc.
corresponding to the incremental change in nodal power injec- Furthermore, the legacy (passive) distribution systems are
tion. But, these methods suffer from the dependency over being transformed into active distribution systems by the
the slack bus, negative loss allocation (cross-subsidies) and widespread deployment of DGs. These active components
over recoveries [17] and therefore needs normalization by alter both magnitude and direction of power flow among dis-
taking certain arbitrary assumptions. Direct loss coefficient tribution feeders so affect power losses. In fact, optimally
method relates losses directly to nodal injections therefore placed DG units can reduce losses in distribution systems
do not require reconciliation [16]. However, the method does besides performance improvement [31]. In several loss alloca-
not seem to be able to do the allocation of the cross-terms tion methods [19], [29], [30], [32], the loss reduction caused
of losses [3]. This led to the development of circuit theory- by DGs is reflected as reduced loss allocation to load points,
based methods. These methods are defined on the basis of though these load points are not actually contributing towards
the system structure, bus impedance matrix and the results of loss reduction. The loss allocation strategy should be such
power flow calculation [18]. Conejo et al. [17] employed bus as to benefit (remunerate) DG owners who have actually
impedance matrix to allocate power loss, but this method can contributed for loss reduction. Moreover, the reactive power
allocate negative losses to some strategically well placed users consumed/injected by the end user is of high significance as it
thus causes cross-subsidies. Fang and Ngan [18] proposed directly affects feeder power losses in distribution systems.
Succinct method for loss allocation using branch-oriented Therefore, the loss allocation method must encourage end
approach. In this method, the allocated losses are obtained users to maintain better power factor. Hence, there is a press-
by the projections of the contributing load current phasors ing need to develop a suitable loss allocation method which
onto the voltage drop phasor of the branch under considera- takes into account the
tion. Reference [19] has pointed out that this method presents 1) fair allocation of cross-terms pertaining to power
a paradox under certain reactive power loads and rectifies the losses
same by assuming reactance free distribution network. Though 2) reactive power consumption/injection by end users
the method of [19] avoids paradox implicitly of the Succinct 3) judicious remuneration to DG owners.
method, but cannot be applied whenever load and distributed In the light of above discussion, this paper proposes a new
generation (DG) are connected at the same node. An Exact circuit theory-based loss allocation method keeping in view
loss allocation method is suggested by Savier and Das [20] by realities of active distribution systems. The proposed method
developing linear relationship between loss allocation and the employs current summation approach in which the cross-terms
power delivered. However, the method [20] is quite analogous of the contributing currents related to load points or DGs
to the Succinct method and therefore suffers from the same are decomposed by proposing a suitable loss allocation factor
drawback. Lim et al. [21] proposed a new method for allocat- which is not based upon heuristic, but is well supported by
ing losses in power system using a loop-based representation providing an analytical treatment. It utilizes the Superposition
of system behaviour, but the method is not applicable to radial principle while allocating remuneration or penalties to DG
distribution systems. Atanasovski and Taleski [23] proposed owners. Salient features of the proposed method are its orig-
energy summation method for loss allocation by assign- inality, simplicity and ability to fairly allocate losses among
ing average node voltages. These voltages are determined network users considering absorption or injection of reactive
once when average loads and generations are applied at all power. The remainder of this paper is organized as follows:
load/generation points. But, the scenario of average load and Section II presents the proposed loss allocation method for
average generation seldom occur. Ghofrani-Jahromi et al. [24] passive and active distribution systems. A detailed investi-
suggested a recursive approach which initiates by assigning gation of the proposed method is carried using 33-bus test
zero power losses to a specific group of nodes. The authors’ distribution system in Section III and a brief discussion
accepted that the method results in over-recovery of losses. is presented in Section IV. Finally, work is concluded in
Several analytical expressions for simultaneous loss alloca- Section V.
tion to DGs and load points are developed in [25]–[27] using
Shapley value and linear circuit theory. However, distribu-
tion loads are usually considered as constant power type so II. P ROPOSED C ROSS -T ERM
these methods are not appropriate for distribution systems. D ECOMPOSITION M ETHOD
Recently, the current summation approach [28] for transmis- The proposed cross-term decomposition method (CTDM)
sion loss allocation is employed in [29] and [30] to allocate considers the branch-oriented approach to allocate losses
losses in distribution systems. In these methods the cross-terms among contributing nodes. The branch losses are composed of
of contributing currents are decomposed using logarithmic square and cross-terms of contributing node currents. Consider
scheme. The authors’ admitted that for fair loss allocation the branch ij of a distribution feeder as shown in Fig. 1.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KANPUR. Downloaded on June 28,2022 at 05:33:30 UTC from IEEE Xplore. Restrictions apply.
 KUMAR et al.: CIRCUIT THEORY-BASED LOSS ALLOCATION METHOD FOR ACTIVE DISTRIBUTION SYSTEMS 1007

∂CT(ij)
CT(ij, n(ij, k)) = I(n(ij, k)) (9)
∂I(n(ij, k))
CT(ij) denotes the cross-term of loss contributed by the current
I(n(ij, k)) in branch ij, however, its derivative term is indepen-
dent of this current itself, as shown in (6). Differentiating (9)
Fig. 1. Single line diagram of system.
with respect to I(n(ij, k)) yields
∂CT(ij, n(ij, k)) ∂CT(ij)
= (10)
The current flowing through the branch ij is ∂I(n(ij, k)) ∂I(n(ij, k))

I(ij) = I(n(ij, k)) (1) From Eqns. (9) and (10)
k∈CN(ij) ∂CT(ij, n(ij, k)) ∂I(n(ij, k))
= (11)
The power loss through the branch ij is CT(ij, n(ij, k)) I(n(ij, k)))

ploss(ij) = ploss(ij, n(ij, k)) Equation (11) reveals that there exist a linear relationship
k∈CN(ij) between CT(ij, n(ij, k)) and I(n(ij, k)). Therefore, to allo-
  cate cross-terms of branch power loss, the loss allocation
= R(ij) (I(ij))2 + Im(I(ij))2 (2) factor (LAF) assigned to the node n(ij, k) is proposed as
⎧ 2 ⎫
⎪
⎨ ⎪
⎬ I(n(ij, k))
k∈CN(ij) I(n(ij, k)) cos φ(n(ij, k)) K(ij, n(ij, k)) = (12)
= R(ij)  2 k∈CN(ij) I(n(ij, k))
⎪
⎩+ ⎪
⎭
k∈CN(ij) I(n(ij, k)) sin φ(n(ij, k)) The cross-terms of power loss to each contributing node
(3) current I(n(ij, k)) in branch ij thus can be allocated as

= R(ij) I 2 (n(ij, k)) CT(ij, n(ij, k)) = K(ij, n(ij, k))CT(ij) (13)
k∈CN(ij)
⎛ ⎞ The loss contributed by kth node in branch ij may be
  I(n(ij, k))I(n(ij, q)) evaluated as
+ R(ij) ⎝× cos{φ(n(ij, k)) ⎠
k∈CN(ij) q=k − φ(n(ij, q))} ploss(ij, n(ij, k)) = R(ij)(I(n(ij, k)))2 + K(ij, n(ij, k))CT(ij)
q∈CN(ij) (14)
(4)
The loss allocated to kth node and the system losses are
= ST(ij) + CT(ij) (say) (5) therefore given by

The first term of (5) clearly provides partial contribution of 

NB
ploss(k) = ploss(ij, n(ij, k)) (15)
each contributing node for the power loss in the branch ij, but
ij=1
the second term needs a bifurcation to allocate power losses
among the contributing nodes. It is noteworthy that the later 
N

term may contribute majority of the power losses in distri- Ploss = ploss(k). (16)
bution systems with long distribution feeders. Reference [28] k=1
demonstrates the comparison of several approaches, namely
proportional, quadratic, geometric, and fast geometric to allo- A. Remuneration/Penalty to DG Owners
cate power losses among contributing nodes corresponding to Using Superposition
cross-terms. However, no explanation is provided for the for- The integration of DG units in distribution system can con-
mulation used, as they are based upon heuristics. Therefore, it tribute significantly in feeder power loss reduction. Therefore,
is important to investigate the dependency of cross-terms on DG owners should be remunerated for their contributions in
contributing node currents. Let us consider, loss reduction, if any. The proposed method employs the
∂CT(ij)   I(n(ij, q)) × cos{φ(n(ij, k))  Superposition principle to evaluate contributing branch cur-
= R(ij)
∂I(n(ij, k)) − φ(n(ij, q))} rents of each DG unit. The current in branch ij without DGs
q=k may be considered as the phasor difference of the branch cur-
q∈CN(ij)
rent with DG and the sum of contributing DG currents in the
(6) branch as shown below
Using (4)-(6)
 I u (ij) = I c (ij) − I DG (ij) (17)
1 ∂CT(ij)
CT(ij) = I(n(ij, k)) Therefore, the following dot product holds good
2 ∂I(n(ij, k))
k∈CN(ij)   
1  I u (ij).I u (ij) = I c (ij) − I DG (ij) . I c (ij) − I DG (ij) (18)
= CT(ij, n(ij, k)) (say) (7)
2 Iu2 (ij)= Ic2 (ij) + I DG (ij).I DG (ij) − 2I c (ij).I DG (ij) (19)
k∈CN(ij)    
I DG (ij).I DG (ij)
where, Iu2 (ij) − Ic2 (ij) R(ij) = R(ij) (20)
 I(n(ij, k))I(n(ij, q))  − 2I c (ij).I DG (ij)
CT(ij, n(ij, k)) = 2R(ij) where,
cos{φ(n(ij, k)) − φ(n(ij, q))}
q =k 
q∈CN(ij) I DG (ij) = I DG (ij, p) (21)
(8) p∈CDG(ij)

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KANPUR. Downloaded on June 28,2022 at 05:33:30 UTC from IEEE Xplore. Restrictions apply.
 1008 IEEE TRANSACTIONS ON SMART GRID, VOL. 10, NO. 1, JANUARY 2019

⎡   ⎤
I DG (ij, p). I DG (ij, p)
⎢ p∈CDG(ij) ⎥
RDG (ij) = ⎢  ⎥R(ij)
p∈CDG(ij)
⎣ − 2I (ij). I DG (ij, p) ⎦
c
p∈CDG(ij)
(22)
⎡  ⎤
2
IDG (ij, p)
⎢ p∈CDG(ij) ⎥
⎢   ⎥
⎢ ⎥
⎢+ I DG (ij, p).I DG (ij, r) ⎥
=⎢
⎢ p∈CDG(ij) r=p
⎥R(ij)
⎥
⎢ ⎥
⎢ 
r∈CDG(ij)
⎥
⎣ − 2I c (ij). I DG (ij, p) ⎦
p∈CDG(ij)

= RDG (ij, p) (23)
p∈CDG(ij)

where, RDG (ij, p) denotes remuneration allocated to the pth

DG for loss reduction in branch ij and is given by
⎡ 2  ⎤
IDG (ij, p) + 2K(ij, p) I DG (ij, p)
⎢ r=p ⎥
RDG (ij, p) = ⎣ ⎦R(ij)
r∈CDG(ij)
× I DG (ij, r) − 2I c (ij).I DG (ij, p)
(24)
where, K(ij, p) is the remuneration allocation factor for the pth
DG owner while bifurcating cross terms of DG contributing
currents in branch ij.
= ST(ij, p) + CT(ij, p) + MT(ij, p) (say) (25)
where,
ST(ij, p) = IDG
2
(ij, p)R(ij) (26)

CT(ij, p) = 2K(ij, p)R(ij) I DG (ij, p).I DG (ij, r) (27)
r=p
r∈CDG(ij) Fig. 2. Flow chart of proposed CTDM.

MT(ij, p) = −2I c (ij).I DG (ij, p)R(ij) (28)

Therefore, remuneration allocated to the pth DG owner is insignificant in contemporary distribution systems having large
given by number of DGs. The insignificant error produced can be pro-
NB NB 
  portionally adjusted among DGs, using (30), on the basis of
ST(ij, p) + CT(ij, p) their respective contribution towards loss reduction without
RDG (ij, p) = (29)
+ MT(ij, p) much affecting the accuracy of the results from practical point
ij=1 ij=1
of view. The adjusted value of remuneration to each DG owner
In (29), the calculations of the first and the third term can be evaluated using (31). The flow chart of the proposed
are straightforward. However, the second term involves cross- CTDM is shown in Fig. 2. It can be observed from the figure
terms related to the contributing DG currents which can be that the proposed method requires only NDG + 2 load flows to
handled in the same way as that of the cross-terms related to allocate losses or remunerate/penalize end users and therefore
contributing load currents. is not computationally demanding.
⎛ ⎞
NB  
ST(ij, p) + MT(ij, p) + CT(ij, p)
εp = ε⎝  ⎠
ij=1
 III. S IMULATION R ESULTS
NDG NB 
p=1 ij=1 ST(ij, p) + MT(ij, p) + CT(ij, p) The 12.66 kV 33-bus test distribution system is considered
(30) to investigate the effectiveness of the proposed loss allocation

NB NB 
  method. The nominal active and reactive power demands for
ST(ij, p) + CT(ij, p)
RDG (p) = RDG (ij, p) = + εp this system are 3.715 MW and 2.3 MVAr, respectively, and
+ MT(ij, p) the detailed line and bus data may be referred from [33]. The
ij=1 ij=1
(31) power loss for this system under nominal loading is found to
be 202.67 kW while the network is operated in base config-
A small error ε may arise while employing superposition uration, i.e., by opening lines 33-37. For load flow backward
to remunerate DG owners if loads are assumed to be constant forward sweep method is used.
power type. It happens due to the variation in node voltages The nominal system loading is considered for simulations.
while applying Superposition. However, this variation will be The proposed CTDM and other established methods, i.e.,

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KANPUR. Downloaded on June 28,2022 at 05:33:30 UTC from IEEE Xplore. Restrictions apply.
 KUMAR et al.: CIRCUIT THEORY-BASED LOSS ALLOCATION METHOD FOR ACTIVE DISTRIBUTION SYSTEMS 1009

Fig. 3. Comparison of loss allocation using CTDM with established methods.

operated at 0.9 power factor while generating reactive power.

The schematic single line diagram of this modified 33-bus test
distribution system is shown in Fig. 4. The figure shows the
network in base configuration with three DGs. The sizing and
siting of these DG units may be referred from Table I. In
the first study DGs are assumed to be active power sources
only. With this DG placement the feeder power losses are
reduced from 202.67 kW to 71.46 kW. Thus a loss reduc-
tion of 131.21 kW is caused by DGs which is remunerated
to DG owners using proposed method. The remuneration may
be in terms of amount proportional remuneration in kWh. The
simulation results obtained and the related calculations for the
contributing branch currents from DGs are explicitly presented
in Table VII of the Appendix. The last two columns of the
table shows a close match between the sum of contributing
branch currents by DGs as obtained using load flow to that
obtained using proposed method. This shows the simplicity
and effectiveness of the proposed method to determine the
branch current contributions by individual DG in active distri-
bution systems. The remuneration allocated to DG owners is
evaluated using proposed method and is presented in Table II.
The table shows the calculated value for total remuneration
of 125.27 kW against the true loss reduction of 131.21 kW
Fig. 4. Single line diagram of modified 33-bus test distribution system. obtained by load flow. This shows an error of −5.95 kW. This
error is divided among DG owners using (30). The adjusted
values of remunerations to DG owners are also shown in
the table.
BCDLA [19], and Exact Method [20] are applied to this sys- In second study, DGs are assumed to supply both active and
tem. It has been observed that all these methods allocate reactive power as given in Table I. For this case, the remu-
total loss of 202.67 kW. The comparison of loss allocation neration allocated to DG owners is presented in Table III.
using these methods is presented in Fig. 3. It can be observed The table shows that the remuneration to all DG owners
from the figure that the proposed method provides compara- increased appreciably. It happened because the reactive power
ble results to other established methods at all nodes except support provided by the DG units result in further loss reduc-
30 to 32. It can also be observed that the proposed method tion, i.e., from 71.46 kW to 19.70 kW. It implies that with
provides relatively more loss allocation than other methods at reactive power injection the remuneration to DG owners is
node 30 having exceptionally poor power factor of 0.32 (lag- increased from 131.21 kW to 182.97 kW, as shown in the
ging) with relatively large load. Similarly, less loss allocation table. Thus proposed method appreciates DG owner to inject
is provided for nodes 31 and 32 having good power factor reactive power into the system. Fig. 5 shows loss allocation
(more than 0.9 lagging) with large values of loads. It may to all load points and the remuneration allocated to DG own-
also be seen that the effect of power factor is pronounced if ers in two different scenarios of power generation. It can be
load is comparatively large. This seems to be more practical observed from the figure that the loss allocated to all load
and judicious approach for loss allocation. points remains unaltered irrespective of the power generated
In order to investigate the applicability of proposed method by DGs. However, the remunerations allocated to DG owners
for active distribution systems, the system is modified by are affected. This implies that the loss allocated to load points
placing three DGs as in [34]. Each DG is assumed to be is not affected by the presence of DGs. This is true because

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KANPUR. Downloaded on June 28,2022 at 05:33:30 UTC from IEEE Xplore. Restrictions apply.
 1010 IEEE TRANSACTIONS ON SMART GRID, VOL. 10, NO. 1, JANUARY 2019

Fig. 5. Loss allocation to load points and DG owners in different scenarios.

TABLE I TABLE V
S ELECTED S IZING AND S ITING OF DG S ACCURACY OF P ROPOSED CTDM W ITH DG S
(ACTIVE AND R EACTIVE P OWER )

TABLE II
R EMUNERATION A LLOCATION TO DG S (ACTIVE P OWER )

TABLE VI
C OMPARISON R ESULTS OF P ROPOSED CTDM W ITH E XISTING M ETHODS

TABLE III
R EMUNERATION A LLOCATION TO DG S (ACTIVE AND R EACTIVE P OWER )

varies from 1-5% for specific load conditions. This shows

TABLE IV that the mean error lies within 2-3%. The error produced
ACCURACY OF CTDM W ITH DG S (ACTIVE P OWER )
seems to be acceptable in the view of nominal error pro-
duced by loss allocation methods which is attributed to
the non-linear relationship between losses and feeder cur-
rents. It can also be observed that DG owners are penal-
ized (negative remuneration) under light load conditions due
to over generation and consequent increase in the feeder
power losses. This shows that proposed CTDM remunerate
or penalize DG owners depending upon their impact on the
net losses.
DGs are causing loss reduction and not the load points. The
figure also shows that all DG owners are remunerated for the B. Validation With Existing Methods
nominal load condition and with power generations as shown
In order to validate the proposed method, it is applied to
in Table I. But, DG owners may be penalized if generation
the same system under identical loading conditions, DG allo-
from DGs cause any increase in power loss. This fact can
cation and their power factors, as in existing methods given
be verified from the next section where the accuracy of the
in [30]. The results of comparison are presented in Table VIII
proposed method is investigated for different load conditions.
of the Appendix. The table shows loss and remuneration
allocated to network users in kW. It can be observed from
A. Accuracy of Proposed CTDM the table that both losses allocated to load points and the
In order to check the accuracy of the proposed CTDM, remuneration allocated to DG owners is different using dif-
the losses/remuneration are calculated by the method for this ferent methods. As a consequence the total allocation to DG
system with load factor of 0.2, 1.0 and 2.0. The results are owners (ADG) and allocation to load points (AL) are quite
compared with their respective true values that are obtained different using different methods, as shown in Table VI. But,
using load flow in Table IV and V. It can be observed that the net loss allocation (NLA), i.e., the net revenue earned
the percentage error using Superposition in proposed method by the utility is identical, i.e., 43.44 kW, using each of

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KANPUR. Downloaded on June 28,2022 at 05:33:30 UTC from IEEE Xplore. Restrictions apply.
 KUMAR et al.: CIRCUIT THEORY-BASED LOSS ALLOCATION METHOD FOR ACTIVE DISTRIBUTION SYSTEMS 1011

TABLE VII
S YSTEM B RANCH C URRENTS FOR D IFFERENT C OMBINATION OF DG S

these methods. It happened because the system losses are proposed CTDM there is no error from utility point of view
43.44 kW with DGs, whereas it is 202.67 kW without DGs. as the error is adjusted among all DG owners.
Therefore, ADG should be 159.67 kW. However, this is found
to be partially diverted to load points by other methods.
This fact can be validated from the table while comparing V. C ONCLUSION
ADG and AL. The loss allocation in active distribution systems is a highly
challenging task on the account of nonlinearity between the
feeder power losses and line flows. This paper proposes a new
circuit theory-based loss allocation method for active dis-
IV. D ISCUSSION tribution systems. Proposed CTDM employs branch-oriented
In the proposed CTDM, Principle of superposition is applied approach to allocate the crossed terms of power loss among
to evaluate contributing branch currents of each DG unit which the contributing network users supported by an analytical
is applicable under identical voltage conditions of all DGs. treatment and employs the principle of Superposition to remu-
However, when there is not adequate voltage support, the sys- nerate/penalize DG owners. The application result shows that
tem node voltages will not be identical which may result in the proposed method is comparable with other established
an error in calculated and actual value of remuneration. It is methods and the slight deviation in results is observed due
important to note that legacy distribution systems are operated to the fact that it also takes into account the effect of power
in radial configuration so there is obvious variation of node factors for relatively larger loads. The noteworthy feature
voltages from source to tail end. However, in contemporary of the proposed method is that it remunerate/penalize DG
distribution system, the DGs are being installed as sources of owner by the equal amount by which DG is contributing
energy at selected node points, consequently there is a kind towards loss reduction/enhancement. Since loads are assumed
of voltage support to the system. Larger the number of DG to be constant power type, a small error may be expected
penetration, larger will be the consequent voltage support. In on account of variation in node voltage profile while remu-
the light of these, it is expected that there will be insignificant nerating or penalizing DG owners. However, these variations
(less than 2%) variation of node voltages and the assumption of are insignificant for contemporary distribution system hav-
identical voltage of all DG nodes holds. It has been found that ing numerous DG units. The salient features of the proposed
when the system load is varied from 20-200% of their nominal method are its originality, simplicity and ability to judi-
value, there is an error in the range of 1-5% in calculated remu- ciously allocate losses and remuneration among network users
nerations and actual remunerations with three DGs. This error on the basis of their net contribution toward feeder power
is proportionally adjusted among DGs, using (30). Therefore, losses.
net error in remuneration to each DG owner is reduced. This
error will be further reduced to insignificant value in con-
temporary distribution systems having large number of DGs A PPENDIX
with consequent voltage support. It is noteworthy that in the See the Tables VII and VIII.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KANPUR. Downloaded on June 28,2022 at 05:33:30 UTC from IEEE Xplore. Restrictions apply.
 1012 IEEE TRANSACTIONS ON SMART GRID, VOL. 10, NO. 1, JANUARY 2019

TABLE VIII
C OMPARISON R ESULTS OF P ROPOSED CTDM W ITH E XISTING M ETHODS [12] O. I. Elgerd, Electric Energy Systems Theory: An Introduction.
New York, NY, USA: McGraw-Hill, 1982.
[13] F. Schweppe, M. Caramanis, R. Tabors, and R. Bohn, Spot Pricing of
Electricity. Boston, MA, USA: Kluwer Acad., 1988.
[14] M. Meisingset and Breidablik, “A method to determine charging prin-
ciples for losses in the Norwegian main grid,” in Proc. 13th PSCC
Trondheim, Trondheim, Norway, Jun./Jul. 1999, pp. 382–387.
[15] F. D. Galiana, A. J. Conejo, and I. Kockar, “Incremental transmission
loss allocation under pool dispatch,” IEEE Trans. Power Syst., vol. 17,
no. 1, pp. 26–33, Feb 2002.
[16] J. Mutale, G. Strbac, S. Curcic, and N. Jenkins, “Allocation of losses
in distribution systems with embedded generation,” IEE Proc. Gener.
Transm. Distrib., vol. 147, no. 1, pp. 7–14, Jan. 2000.
[17] A. J. Conejo, F. D. Galiana, and I. Kochar, “Z-bus loss allocation,” IEEE
Trans. Power Syst., vol. 16, no. 1, pp. 105–110, Feb. 2001.
[18] W. L. Fang and H. W. Ngan, “Succinct method for allocation of network
losses,” IEE Proc. Gener. Transm. Distrib., vol. 149, no. 2, pp. 171–174,
Mar. 2002.
[19] E. Carpaneto, G. Chicco, and J. S. Akilimali, “Branch current decom-
position method for loss allocation in radial distribution systems with
distributed generation,” IEEE Trans. Power Syst., vol. 21, no. 3,
pp. 1170–1179, Aug. 2006.
[20] J. S. Savier and D. Das, “Energy loss allocation in radial distribution
systems: A comparison of practical algorithms,” IEEE Trans. Power
Del., vol. 24, no. 1, pp. 260–267, Jan. 2009.
[21] V. S. C. Lim, J. D. F. McDonald, and T. K. Saha, “Development of
a new loss allocation method for a hybrid electricity market using graph
theory,” Elect. Power Syst. Res., vol. 79, no. 2, pp. 301–310, Feb. 2009.
[22] N. Jenkins, R. Allan, P. Crossley, D. Kirschen, and G. Strbac, Embedded
Generation (IEE Power and Energy Series 31). London, U.K.: Inst.
Elect. Eng., 2000.
[23] M. Atanasovski and R. Taleski, “Energy summation method for loss
allocation in radial distribution networks with DG,” IEEE Trans. Power
Syst., vol. 27, no. 3, pp. 1433–1440, Aug. 2012.
[24] Z. Ghofrani-Jahromi, Z. Mahmoodzadeh, and M. Ehsan, “Distribution
loss allocation for radial systems including DGs,” IEEE Trans. Power
Del., vol. 29, no. 1, pp. 72–80, Feb. 2014.
[25] S. Sharma and A. R. Abhyankar, “Loss allocation for weakly meshed
distribution system using analytical formulation of Shapley value,” IEEE
Trans. Power Syst., vol. 32, no. 2, pp. 1369–1377, Mar. 2017.
[26] J.-C. Peng, H. Jiang, and Y.-H. Song, “A weakly conditioned imputation
of an impedance-branch dissipation power,” IEEE Trans. Power Syst.,
vol. 22, no. 4, pp. 2124–2133, Nov. 2007.
R EFERENCES [27] Y. P. Molina, R. B. Prada, and O. R. Saavedra, “Complex losses
[1] J. J. Gonzalez and P. Basagoiti, “Spanish power exchange market and allocation to generators and loads based on circuit theory and
information system design concepts, and operating experience,” in Proc. Aumann–Shapley method,” IEEE Trans. Power Syst., vol. 25, no. 4,
IEEE PICA Conf., Santa Clara, CA, USA, May 1999, pp. 245–252. pp. 1928–1936, Nov. 2010.
[28] A. G. Expósito, J. M. R. Santos, T. G. Garćia, and E. A. R. Velasco,
[2] J. W. Bialek, S. Ziemianek, and N. Abi-Samra, “Tracking-based loss
“Fair allocation of transmission power losses,” IEEE Trans. Power Syst.,
allocation and economic dispatch,” in Proc. 13th PSCC, Trondheim,
vol. 15, no. 1, pp. 184–188, Feb. 2000.
Norway, Jul. 1999, pp. 375–381.
[29] K. M. Jagtap and D. K. Khatod, “Loss allocation in radial distribution
[3] P. M. Costa and M. A. Matos, “Loss allocation in distribution networks
networks with different load models and distributed generations,” IET
with embedded generation,” IEEE Trans. Power Syst., vol. 19, no. 1,
Gener. Transm. Distrib., vol. 9, no. 12, pp. 1275–1291, Sep. 2015.
pp. 384–389, Feb. 2004.
[30] K. M. Jagtap and D. K. Khatod, “Novel approach for loss allocation
[4] J. S. Savier and D. Das, “Loss allocation to consumers before and of distribution networks with DGs,” Elect. Power Syst. Res., vol. 143,
after reconfiguration of radial distribution networks,” Int. J. Elect. Power pp. 303–311, Feb. 2017.
Energy Syst., vol. 33, no. 3, pp. 540–549, Mar. 2011. [31] P. S. Georgilakis and N. D. Hatziargyriou, “Optimal distributed gener-
[5] J. W. Bialek, “Identification of source-sink connections in transmission ation placement in power distribution networks: Models, methods, and
networks,” in Proc. 4th Int. IEE Conf. Power Syst. Control Manag., future research,” IEEE Trans. Power Syst., vol. 28, no. 3, pp. 3420–3428,
London, U.K., Apr. 1996, pp. 200–204. Aug. 2013.
[6] J. W. Bialek, “Tracing the flow of electricity,” IEE Proc. Gener. Transm. [32] M. Atanasovski and R. Taleski, “Power summation method for loss allo-
Distrib., vol. 143, no. 4, pp. 313–320, Jul. 1996. cation in radial distribution networks with DG,” IEEE Trans. Power
[7] D. Kirschen, R. Allan, and G. Strbac, “Contributions of individual gen- Syst., vol. 26, no. 4, pp. 2491–2499, Nov. 2011.
erators to loads and flows,” IEEE Trans. Power Syst., vol. 12, no. 1, [33] M. E. Baran and F. F. Wu, “Network reconfiguration in distribution
pp. 52–60, Feb. 1997. systems for loss reduction and load balancing,” IEEE Trans. Power Del.,
[8] J. W. Bialek, “Topological generation and load distribution factors for vol. 4, no. 2, pp. 1401–1407, Apr. 1989.
supplement charge allocation in transmission open access,” IEEE Trans. [34] P. Kumar, N. Gupta, A. Swarnkar, and K. R. Niazi, “Discrete particle
Power Syst., vol. 12, no. 3, pp. 1185–1193, Aug. 1997. swarm optimization for optimal DG placement in distribution networks,”
[9] G. Strbac, D. Kirschen, and S. Ahmed, “Allocating transmission system in Proc. 18th NPSC, Guwahati, India, Dec. 2014, pp. 1–6.
usage on the basis of traceable contributions of generators and loads to
flows,” IEEE Trans. Power Syst., vol. 13, no. 2, pp. 527–534, May 1998.
[10] D. Kirschen and G. Strbac, “Tracing active and reactive power between
generators and loads using real and imaginary currents,” IEEE Trans.
Power Syst., vol. 14, no. 4, pp. 1312–1319, Nov. 1999.
[11] Guidance Note for Calculation of Loss Factors for Embedded
Generation in Settlement, Electricity Pool England Wales, London, U.K.,
Apr. 1992. Authors’ photographs and biographies not available at the time of publication.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KANPUR. Downloaded on June 28,2022 at 05:33:30 UTC from IEEE Xplore. Restrictions apply.