You are on page 1of 8

260 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO.

1, JANUARY 2009

Energy Loss Allocation in Radial Distribution


Systems: A Comparison of Practical Algorithms
J. S. Savier and Debapriya Das

Abstract—This paper presents allocation of energy losses to reactive component of load current at node
consumers connected to radial distribution networks in a deregu-
lated environment. A detailed comparison of the proposed “Exact
Method” with three alternative algorithms, namely, pro rata, real power loss of branch- ;
quadratic allocation and proportional allocation are presented.
Pro rata procedure is based on the load demand of each consumer, resistance of branch- ;
quadratic and proportional allocations are based on identifying
the real and reactive parts of current in each branch and the losses real part of complex quantity, ;
are allocated to each consumer, and the proposed “Exact Method” real power loss of branch- allocated
is based on the actual contribution of real power loss by each
consumer. A case study based on a 30-node distribution system is
to consumer connected at node
provided. ;
Index Terms—Deregulation, energy loss allocation, radial distri- real power loss supported by consumer
bution systems. ;
duration of load level in hours;
I. NOMENCLATURE total power loss allocated to consumer at
load level.
node number;
kVA load demand of consumer at node ; II. INTRODUCTION
total real power loss in the system; EREGULATION of the power industry was intended to
total number of nodes in the system; D introduce competition and to bring down prices, improve
efficiency, and generate an overall improvement in the industry.
load current at node; It was primarily aimed at demolishing the monopolistic, protec-
real power load at node; tive and vertically integrated structure of the industry and pro-
viding a chance for the new entrants to make their contributions
reactive power load at node; to the business in general. Unlike generation and sale of elec-
voltage at node; trical energy, activities of transmission and distribution are gen-
erally considered as a natural monopoly. The cost of transmis-
branch number, ;
sion and distribution activities needs to be allocated to the users
current of branch- ; of these networks. Allocation can be done through network use
total number of nodes beyond branch- ; tariffs, with a focus on the true impact they have on these costs.
Among others, distribution energy losses are one of the costs
nodes beyond branch- , for to be allocated. The main difficulty faced in allocating losses is
; the nonlinearity between the losses and delivered power which
loss allocation factors for the consumer at complicates the impact of each user on network losses [1].
node ; Different techniques have been published in the literature for
allocation of real power losses, most of them dedicated to trans-
loss allocation factors for the consumer at mission networks and can be classified into three broad cate-
node ; gories—pro rata procedures, marginal procedures and propor-
load current of consumer at node ; tional sharing procedures. Pro rata procedure [2], [3] is the sim-
plest one, in which, the total losses are allocated to loads or gen-
real component of load current at node
erators based on the load active power demand or bus genera-
;
tion. In marginal procedures [4]–[7], losses are assigned to gen-
erators and demands through the so-called incremental trans-
Manuscript received April 30, 2007; revised August 13, 2008. Current version mission loss (ITL) coefficients. In proportional sharing proce-
published December 24, 2008. Paper no. TPWRD-00243-2007. dures [8]–[13], losses are allocated to the generators and con-
The authors are with the Electrical Engineering Department, Indian Insti- sumers by using the results of a converged power flow plus a
tute of Technology, Kharagpur 721302, India (e-mail: savier_js@yahoo.com;
ddas@ee.iitkgp.ernet.in). linear proportional sharing principle. Conejo et al. [14] have
Digital Object Identifier 10.1109/TPWRD.2008.2007018 presented a new procedure for allocating transmission losses to
0885-8977/$25.00 © 2008 IEEE
SAVIER AND DAS: ENERGY LOSS ALLOCATION IN RADIAL DISTRIBUTION SYSTEMS: A COMPARISON OF PRACTICAL ALGORITHMS 261

generation and loads based on the network Z-bus matrix. Conejo A. Pro Rata (PR) Allocation Method
et al. [15] have also presented a comparison of four different The pro rata method proportionally allocates losses to the
practical algorithms for transmission loss allocation. Fang and consumers based on the kW load demand (in the case of trans-
Ngan [16] proposed a succinct method for allocation of network mission systems) [15]. In order to take into account reactive load
losses which takes into consideration the influence of both ac- of each consumer, instead of considering real power demand of
tive and reactive power injected into grids. Costa and Matos [17] the consumer, kVA load demand is considered for radial distri-
have addressed the allocation of losses in distribution networks bution systems.
with embedded generation by considering quadratic loss alloca- The total kVA demand of the system can be written as
tion technique.
As there is no unique or ideal procedure available, any loss
(1)
allocation algorithm should have most of the desirable proper-
ties stated below [15]:
1) to be consistent with the results of a power flow; Note that substation is marked as node 1, i.e., .
2) to depend upon the amount of energy either produced or Hence, power loss allocated to consumer at node is given
consumed; by
3) to depend on the relative location in the distribution net- (2)
work;
4) to avoid volatility;
(3)
5) to be easy to understand;
6) to be easy to implement. It may be noted that the demand loss allocation factors
In the light of the above developments, this work proposes a are identical for all buses. Hence, pro rata procedure is simple
simple loss allocation method, called the “Exact Method”, for to understand and implement but they ignore the relative loca-
energy loss allocation to consumers in radial distribution sys- tion of the consumers within the network. That is, two identical
tems. In the proposed “Exact Method”, energy losses are al- demands located respectively near substation and far away from
located to each consumer by finding the actual contribution of the substation are equally treated, and this is unfair to the load
power loss at different load levels by that consumer and no as- located near the substation.
sumptions regarding the sharing of power losses are used as op-
posed to other algorithms available in the literature. Energy loss B. Quadratic Allocation Method
allocation to each consumer obtained with the proposed method In quadratic loss allocation scheme, since the power losses
is compared with three different practical approaches for loss grow quadratically with power flows, the following constraint
allocation, namely, pro rata, quadratic allocation, and propor- is imposed [1]:
tional allocation.
(4)
III. DISTRIBUTION LOSS ALLOCATION METHODS
and
In radial distribution systems, power is fed at substations and
the power flows from substation to down stream. Energy loss (5)
allocation to each consumer is made based on the real power loss
allocated at different load levels of the load duration curve. Real Based on this principle, the power loss of the branch- of
power loss allocation methods for radial distribution systems the network allocated to consumer at node , beyond
considered in this study are first presented and then the energy branch- , for are (Derivation given in
loss allocation to different consumers. Appendix A) as shown in (6), found at the bottom of the page.

(6)
262 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 1, JANUARY 2009

The global value of losses to be supported by consumer re- In this case also, the global value of losses to be supported by
sults from the sum of the losses allocated to it in each branch- consumer results from the sum of the losses allocated to it in
of the network, i.e., each branch- of the network, i.e.,

(7)
(13)

Note that node 1 is substation. Equation (7) indicates that


each consumer has allocated losses only at branches to which Note that node 1 is substation.
power-flow contributes.
D. Proposed Exact Method
C. Proportional Allocation Method
The proposed Exact Method uses the results of a converged
The proportional loss-allocation scheme is the most intuitive load flow based on the identification of nodes and branches pro-
and straightforward approach results which from the propor- posed in [18].
tionality assumption [1], i.e., In a radial distribution system, current through any branch-
can be written as:
(8)
(14)
and

(9) Real power loss of branch- with sending end and receiving
end voltages and is given by
The power loss of the branch- of the network allocated to
consumers beyond branch- , for can be (15)
derived similar to that of quadratic loss allocation scheme which
is given in (10), shown at the bottom of the page, where
i.e., see (16), at the bottom of the page, and (17)–(19), shown
at the bottom of the next page.
(11) Hence

and

(12)
(20)

(10)

(16)
SAVIER AND DAS: ENERGY LOSS ALLOCATION IN RADIAL DISTRIBUTION SYSTEMS: A COMPARISON OF PRACTICAL ALGORITHMS 263

Using (20), real power loss in branch- can be allocated to


consumers beyond branch- . Real power loss of branch- al-
located to consumer connected to node is given by

(21)

The global value of losses to be supported by consumer con-


nected to node at load level results from the sum of the
losses allocated to it in each branch- of the network, i.e.,

Fig. 1. Distribution system with 30 nodes.

(22) TABLE I
LOAD DURATION CURVE
Note that node 1 is substation.
Total cost of energy losses allocated to consumer for
the load duration curve considered is determined using the
following relation:

(23) closer to the substation is less as compared to those consumers


electrically away from the substation. For example, consumers
at nodes 10 and 28 are having the same load demand. Hence,
In the present study, the cost of energy is assumed as loss allocation based on pro rata procedure allocates same
. losses to both the consumers, i.e., 4.3624 kW, considering that
the load level is 1.0. Whereas, loss allocation to consumers at
IV. CASE STUDY nodes 10 and 28 based on quadratic, proportional, and exact
A practical, 11-kV distribution system having 30 nodes as method are different. Quadratic loss allocation scheme allo-
shown in Fig. 1 is considered to compare the four distribution cated 3.3636 and 3.6851 kW to the consumers at nodes 10 and
loss allocation methods. The line and load data for this system 28, respectively, proportional loss allocation scheme allocated
are given in Appendix B. The loads considered are either indus- 3.7022 and 3.9995 kW to the consumers at nodes 10 and 28,
trial or commercial type. The discretized load duration curve respectively, and exact method allocated 4.2312 kW and 4.4907
data for the system is given in Table I. The total real power kW to the consumers at nodes 10 and 28, respectively. It can be
loss of the system, during the load levels of 1.0 (High), 0.7 seen that all the methods except pro rata method allocated less
(Medium), and 0.3 (Low) are 146.09, 67.61, and 11.59 kW, re- power loss to consumer at node 10, which is electrically closer
spectively. to the substation, as compared to losses allocated to consumer
The real power loss allocated to each consumer with the at node 28, which is electrically away from the substation.
four approaches at different load levels are shown in Figs. 2–4. Energy loss cost allocated to consumers at different nodes,
Power flow data and results are used as input data for the assuming cost of energy as , with the four
four loss allocation algorithms. The real power loss allocation different approaches is shown in Fig. 5. It can be seen from
based on pro rata procedure allocates losses irrespective of Fig. 5 that, cost of energy loss allocated to consumers with pro
the geographical location of consumers. Hence, consumers rata method at node nos. 10 and 28 is the same, i.e., $ 1101.4.
having same load demands are allocated same losses, even But the energy loss cost allocated with all the other methods to
though the power loss contribution of the consumer electrically these two consumers (consumers at node nos. 10 and 28) are

(17)

(18)

(19)
264 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 1, JANUARY 2009

Fig. 2. Real power loss allocated at load level = 1:0 . Fig. 5. Energy loss cost allocated to consumers at different nodes.

Quadratic and proportional loss allocation techniques make


use of assumptions to obtain power loss of each branch allocated
to consumers beyond that branch. The proposed “Exact method”
allocates branch losses to different consumers based on actual
contribution of the branch power losses by each consumer beyond
that branch. In the present work, a load flow algorithm developed
in [18] for solving the radial distribution network has been used.

V. CONCLUSIONS
In the present work, a comparison of energy loss allocation in
a radial distribution network, based on three different methods
namely, pro rata, quadratic, and proportional methods with the
proposed “Exact Method” has been presented. From the case
study, it can be seen that even though pro rata procedure is
Fig. 3. Real power loss allocated at load level = 0:7 .
simple and easy to implement, power loss, and energy loss cost
allocated to consumers having same load demands are the same,
which is an injustice to the consumer electrically nearer to the
substation. Quadratic and proportional loss allocation schemes
are based on branch current flow and these two methods allocate
branch power loss to only those consumers beyond that branch.
Quadratic loss allocation scheme makes the assumption that the
loss allocation factors of a particular consumer is proportional to
the square of real/reactive load current of that consumer whereas
proportional loss allocation scheme assumes that the loss allo-
cation factors are proportional to real/reactive load current. In
the proposed ‘exact method’, losses are allocated to consumers
without making any assumptions and can be implemented easily.
Distribution systems are almost universally socialized in con-
trast with transmission losses that can be attributed to individual
customers and generators. Hence, distribution companies may
adopt a tariff structure in which in addition to the fixed charge
($ per month), capacity charge ($ per kW per month), actual
Fig. 4. Real power loss allocated at load level = 0:3 . demand charge ($ per kW per month) and energy volume charge
($ per kWh), a portion of the actual energy loss caused by the con-
sumer may be charged (say, 50% of the total energy loss caused
different. Quadratic, proportional and proposed methods allo- by a consumer). The remaining energy loss may be allocated
cated $ 851.92, $ 937.11, and $ 1070.75, respectively, towards based on some other criteria (say, based on installed capacity).
energy loss cost to the consumer at node 10 whereas it allocated The proposed “Exact Method” for loss allocation can also be
$ 931.46, $ 1010.63, and $ 1134.94, respectively, to the con- used as a guide in distribution planning. While planning a distri-
sumer at node 28. bution system, engineers can visualize actual real power losses
SAVIER AND DAS: ENERGY LOSS ALLOCATION IN RADIAL DISTRIBUTION SYSTEMS: A COMPARISON OF PRACTICAL ALGORITHMS 265

The load current at any node is given as


(A1)

Consider branch-8 of Fig. 6. Number of node beyond


branch-8 is one and this is node 9. Therefore, current through
branch-8 is
(A2)
Similarly consider branch-7. Total number of nodes beyond
branch-7 is three and these nodes are 8, 9, and 12. Therefore,
Fig. 6. Sample distribution network with 12 nodes. current through branch-7 is

TABLE II
(A3)
BRANCH NUMBER, SENDING END NODE, RECEIVING END NODE, AND NODES
BEYOND DIFFERENT BRANCHES From (A2)–(A3), it is clear that, if we identify the nodes be-
yond all the branches and if the load currents are known, then
it is extremely easy to compute the branch currents. In [18], an
algorithm for identifying the nodes beyond all the branches has
been proposed. A load flow algorithm based on the identification
of these nodes beyond all the branches has also been proposed
in [18]. In the present work, this load flow algorithm is used to
compute the load currents and branch currents.
General expression of branch current through branch- is
given by [18]:

(A4)

In Table II, total number of nodes (consumers) be-


yond branch- and nodes (consumers) beyond
branch- are given for Fig. 6
allocated to consumers having same load demands. The distribu- for the purpose of explanation.
tion system can thus be designed in such a way that the real power Let us define
losses allocated to consumers having same load demands, located (A5)
at different parts of the distribution system, will not differ much.
From (A4) and (A5), we obtain
APPENDIX A
QUADRATIC LOSS ALLOCATION METHOD (A6)

For the purpose of explanation of the quadratic loss-alloca-


Magnitude of current in branch- is expressed as shown in
tion method, consider a sample distribution network as shown
(A7) and (A8), shown at the bottom of the page.
in Fig. 6. In Fig. 6, branch numbers are shown in . The
Power loss of branch- can be written as
branch number, sending end, and receiving end nodes are given
in Table II. (A9)

(A7)

(A8)
266 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 1, JANUARY 2009

From (A8) and (A9), we get (A10), found at the bottom of Equations (A11) and (A12) can be simply written as:
the page.
The development of expression given by (A10), the consumer (A13)
(consumers beyond branch- for
(A14)
) has the impact on the terms and
of the power loss of branch- , which are
where and are the variable loss allo-
exclusively due to consumers for .
cation factors for the consumer at node .
The terms In quadratic loss allocation scheme, since the power losses
are the grow quadratically with power flows, the following constraint
results of the simultaneous influence of consumers is imposed [1].
and the remaining consumers in the components of
the current in the branch- . These crossed terms must also be
allocated. The allocation of crossed terms has been done in a (A15)
quadratic way, as mentioned in [1].
The crossed terms and and
are split into two compo-
nents as described here. Let us define (A16)

From (A13) and (A15), we get,

(A11) (A17)
Similarly, from (A14) and (A16), we get,
and

(A18)
Based on this principle, the power loss of the branch- of
the network allocated to consumers beyond branch- , for
(A12) are shown in (A19), at the bottom of the page.

(A10)

(A19)
SAVIER AND DAS: ENERGY LOSS ALLOCATION IN RADIAL DISTRIBUTION SYSTEMS: A COMPARISON OF PRACTICAL ALGORITHMS 267

The global value of losses to be supported by consumer re- [2] J. J. Gonzalez and P. Basagoiti, “Spanish power exchange market and
sults from the sum of the losses allocated to it in each branch- information system. Design concepts, and operating experience,” in
Proc. 1999 IEEE Power Industry Computer Applications Conf., Santa
of the network, i.e., Clara, CA, May 1999, pp. 245–252.
[3] J. W. Bialek, S. Ziemianek, and N. Abi-Samra, “Tracking-based loss
allocation and economic dispatch,” in Proc. 13th Power Systems Com-
putation Conf., Trondheim, Norway, Jul. 1999, pp. 375–381.
[4] O. I. Elgerd, Electric Energy Systems Theory: An Introduction. New
(A20) York: McGraw-Hill, 1982.
[5] F. Schweppe, M. Caramanis, R. Tabors, and R. Bohn, Spot Pricing of
Note that node 1 is substation. Equation (A20) indicates that Electricity. Boston, MA: Kluwer , 1988.
each consumer has allocated losses only at branches to which [6] M. Meisingset and Ö Breidablik, “A method to determine charging
power flow contributes. principles for losses in the Norwegian main grid,” in Proc. 13th PSCC,
Trondheim, Norway, Jun./Jul. 1999, pp. 382–387.
[7] F. D. Galiana, A. J. Conejo, and I. Kockar, “Incremental transmission
APPENDIX B loss allocation under pool dispatch,” IEEE Trans. Power Syst., vol. 17,
no. 1, pp. 26–33, Feb. 2002.
TABLE III [8] J. W. Bialek, “Identification of source-ink connections in transmission
LINE AND LOAD DATA OF 30 NODE RADIAL DISTRIBUTION SYSTEM networks,” in Proc. 4th Int. IEE Conf. Power System Control and Man-
agement, London, U.K, Apr. 16–18, 1996, pp. 200–204.
[9] J. W. Bialek, “Tracing the flow of electricity,” Proc. Inst. Elect. Eng.,
Gen. Transm. Dist., vol. 143, no. 4, 1996.
[10] D. Kirschen, R. Allan, and G. Strbac, “Contributions of individual gen-
erators to loads and flows,” IEEE Trans. Power Syst., vol. 12, no. 1, pp.
52–60, Feb. 1997.
[11] J. W. Bialek, “Topological generation and load distribution factors
for supplement charge allocation in transmission open access,” IEEE
Trans. Power Syst., vol. 12, no. 3, pp. 1185–1193, Aug. 1997.
[12] G. Strbac, D. Kirschen, and S. Ahmed, “Allocating transmission system
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.
[13] 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.
[14] A. J. Conejo, F. D. Galiana, and I. I. Kockar, “Z-bus loss allocation,”
IEEE Trans. Power Syst., vol. 16, no. 1, pp. 105–110, Feb. 2001.
[15] A. J. Conejo, J. M. Arroyo, N. Alguacil, and A. L. Guijarro, “Transmis-
sion loss allocation: A comparison of different practical algorithms,”
IEEE Trans. Power Syst., vol. 17, no. 3, pp. 571–576, Aug. 2002.
[16] W. L. Fang and H. W. Ngan, “Succinct method for allocation of net-
work losses,” IEE Proc.-Gener. Transm. Distrib., vol. 149, no. 2, pp.
171–174, Mar. 2002.
[17] P. M. Costa and M. A. Matos, “Loss allocation in distribution networks
with embedded generation,” IEEE Trans. Power Syst., vol. 19, no. 1, pp.
384–389, Feb. 2004.
[18] S. Ghosh and D. Das, “Method of load flow solution of radial distribu-
tion networks,” Proc. IEE Gener. Trans. Distrib., vol. 146, no. 6, pp.
641–648, Nov. 1999.

J. S. Savier received the B.Tech. degree from the University of Kerala, Thiru-
vananthapuram, India, in 1996 and the M.E degree from the Indian Institute of
Science, Bangalore, in 2002. He is currently pursuing the Ph.D. degree at the
Indian Institute of Technology, Kharagpur, India.

Debapriya Das received the B. E degree from Calcutta University, Calcutta,


India, in 1982, the M. Tech degree from the Indian Institute of Technology,
Kharagpur, in 1984, and the Ph.D. degree from the Indian Institute of Tech-
REFERENCES nology, Delhi.
[1] A. Gómez, J. M. Riquelme, T. González, and E. Ruiz, “Fair allocation Currently, he is a Professor at the Indian Institute of Technology, Kharagpur.
of transmission power losses,” IEEE Trans. Power Syst., vol. 15, no. 1, His research interests are electric power distribution system and power system
pp. 184–188, Feb. 2000. operation and control.

You might also like