You are on page 1of 5

GIS Based Distribution System Spatial Load

Forecasting and the Optimal Planning of


Substation Location and Capacity
ZHOU Quan, SUN Caixin, CHEN Guoqing, DENG Qun and LIAO Ruijin

Abstract-Geographic information systems(G1S) spatial searching, spatial analysis and the selected substation
platform is used in spatial load forecasting (SLF) and the location retrieving, puts forward a new algorithm which can
mathematical models of the substation location and solve the multi-period optimal selecting problem of substation
capacity optimal problems are established in this paper. based on network flow theory. This method can decide the
Based on the theory of network flow, a new multi-period location and capacity of reasonable substation and its running
optimal selecting algorithm of substation is proposed, time to meet the needs of load development of planning period,
which could determine the reasonable location, capacity and minimize the expense of investment and running
and time that put into running of the substation. The real maintenance.
instance shows that this algorithm could solve the multi
period optimal selecting problem of substation in 11. GIs PLATFORM
distribution system planning (DSP) successfully, and has
the practical value. GIS is a crossing subject and technique among information
science, spatial science and geographic science. Since the
Index Terms-- Geographic information system, Spatial load 1980s, GIS has extended into many applying field, and its
forecasting, Substation, Distribution system planning, Optimal definition has enlarged, in a conclusion, the information
planning, Network flow.
system related geographic location all can be called GIs.
SLF and DSP both involve a great amount of special
1. INTRODUCTION
information. DSP must have a platform, i.e. GIS platform, to

W ith the fast development of city distribution system, it’s correlate the characteristic of distribution planning. GIS
an important subject for power system planning to platform not only provides an environment of distributing
extend distribution system planning (DSP). DSP is a complex devices’ spatial data (e.g. location, shape and connecting
systemic project. The prominent properties of its complexity relation) correlating with the properties’ data, but also supplies
are large-scale, uncertainty, much un-subtle factor and multi- the analytical tools (devices’ connection analysis, min-short
domains. DSP requires not only a great deal of the’ city path analysis) which allow the distribution system exploring
development history data, but also analyses totally the present staff to solve the professional problem on high point [6].
city network and the future city development. It is The mode of planning will change by introducing GIS
dissatisfactory for modern city distribution system to depend platform into DSP. Firstly, GIS is convenient for the
on the experience of planning staff to plan future power system. management of planning data and improve the productivity of
SLF is the basis of DSP and automatic planning. To plan the planning staff. Meanwhile the planning process is more direct
decision- variables of substation location and capacity, the and more interactive. Secondly, GIS conveniently assists
type of wire and its path, the equipment of switcher devices planning staff to decide location of the selected substation and
and their running time, the future load must be established in the corridor of aerial-wire and the path of cable, which may
each area of power distribution domain. SLF not only forecasts shorten the searching scopes of optimal planning procedure
the quantities of future load, but also provides the location of and let the planning result more accordant with the practice.
the load increment, i.e. the spatial distribution of hture load Thirdly, the introduction of GIS platform can synchronize the
[11-[31. DSP with the city planning.
SLF involves a great amount of spatial information. GIS can
provide a platform of collecting, processing data and 111. SPATIAL LOAD FORECASTING
forecasting results for SLF. Applying GIS to SLF can decrease
Load forecasting is the basic work of the distribution system
the quantities of data collecting, so it is practical and necessary
planning, the precision of which directly affects the quality of
for SLF.
the whole planning work. Load forecasting is consist of
This paper, based on SLF and applied GIS’s functions of
classified power and the forecasting of the whole power,
classified load and the forecasting of the entire load, load
ZHOU Quan, SUN Caixin, CHEN Guoqing, DENG Qun and LlAO curve forecasting and load distribution forecasting. The based
Ruijin are with the key Laboratory of High Voltage Engineering and Electrical and pivotal work is the load distribution planning, which is
New Technology, ChongQing University, Ministry of Education. China, also called spatial load forecasting.
400044 (e-mail. hvlab@cqu.edu.cn)

0-7803-7459-2/02/$17.00 02002 IEEE - 885 -


The primary work of the spatial load forecasting is that the C. Gathering the spatial information
demand and distribution of the load in the power supply area Gathering the spatial information is carried out on GIS
should be forecasted. So the power supply area must be platform. To begin with, the city map, which has been
divided into several sections. Thus the spatial load forecasting digitalized, is divided into some layers, such as traffic and road
of the whole area come down to the load forecasting of the layer, resident layer, school layer, commercial center layer,
section. This forecasting method is generally adopted, for it industrial park. To each section, the area of each land style in
have the virtue of easy data-collection, being prone to the existence and the blank area used in the future development
variety of town planning project and strong agility [4],[5]. are analyzed. Three aspects reflect the section characteristic:
The steps of the spatial load distribution: @ The condition of the section itself, e.g. the flatness, and
A . The whole capacify load forecasting whether the town planning restrict the development; @ The
Many methods are available to the whole capacity load distance index of the section, e.g. the distance between the
forecasting, and the precisions of these methods are also section and the center of town and the road and the school; @I
relatively high. And that the programming connecting-network the environment factors, e.g. the commercial status around the
( the transmit electricity system and the sub-transmit electricity section and number of the resident. All of these spatial
system) is based on the whole load forecasting. So the whole characteristics can be picked up from GIS system.
load forecasting should be used in the spatial load forecasting D. Classified landforecasting
directly, and the whole load forecasting is distributed to each
In development of the city, each industry and departments
power supply section, and then the forecasting of the future
are all harmonious and facilitated each other, and a fixed
load distributing of the distribution system is worked out.
proportion is in existence among the loads. It is the objective
B. ClassiJLingthe land and partition of the section of classified land forecasting that the capacity of the land-
The consumer needed to be divided into several types increasing is educed from the whole capacity forecasting using
according to the different characteristic of electric-use and this proportion relationship. The proportion relationship can
land-use in the spatial load forecasting. It is generally divided be attained from statistical data. The available proportion
into 4-20 styles. TABLE I is generally used. relation can be used directly.
According to the land-use forecasting the electric-use
TABLE I THESTYLES OF THELAND-USED forecasting can be attained, and then the spatial load
Great styles Small styles distributing of the distribution system. From above, the steps
Great, inediuin and small enterprises, Storage industry, of the spatial load forecasting method are showed in
Industry
Municioal establishinent
Colnmerce
Retail trade, Show business, Writing buildings,
Guesthouse, high buildings
I-[
Resident I House, Villa $.
Civicism I Official business. Investigate organization Distributingsection and gathering

Both the load characteristic and the land-use demand of


the industry and resident are different. The characteristic of Land-use simulation model
electric-use of each land-use style is showed on the curve of 1 .demand of land-use analysis
representative load density, the variety of which can be
forecasted according to the terminal load forecasting method. I I
Consequently the spatial load forecasting can be transformed
Classified land
into land distribution forecasting. The spatial load distribution
forecasting
can be simply worked out, as long as the spatial land-use
distribution is forecasted.
The distribution of small section is the essential step, the
purpose of which is the forecasting the location of the load-
increased and offering the spatial information for the I
distribution system planning. The smaller the section is Load distribution
compartmentalized, the higher spatial discrimination the load
forecasting will has, the distribution system planning will be
Fig. 1 Overall structure of load-use based SLF method
more particularity. Presently the most method of spatial load
forecasting is dividing the power supply section into several
Iv. OPTIMIZPJG PANNING OF SUBSTATION’S LOCATION AND
equal grilles. This distribution method is propitious to the
CAPACITY
realization of the spatial load forecasting method and the
generalization and standardization. According to the SLF result, the load density and points,
who consume power mass, can be confirmed in every district
during every planning stage. Furthermore, how to properly

- 886 -
determine the geographic location, capacity and the time that J,-@her of the load point substation i supplies the
put into running of every substation in order to satisfy the need power for
of load development in the planning period and minimize the d,-the distance of line from substation i to load pointj
fee of invest and running maintain. Because the geographic &--the capacity of the No. i substation
location, power-supply scope, capacity and the time that put e(SJ-utilization rate of the No. i substation
into running of every substation is unknown and they influence
cos +-power factor
and restrict one another, there are a mass of combination
r -fee coefficient of network wastage
scheme .The whole problem is rather complicated plus other
factors , such as landform and traffic [7],[8].
The substation location and capacity optimal problems are
,.?
,=I , c l ,
d,, in objective function approximately

divided into objective year and middle year, which could be represented the network wastage and investment , when
solved step by step. Using the distribution network planning a,@'') are approximately equal. We can adjust the area of
GIS developed by ourselves, we can determine the geographic
every districts to make the load of every districts
location, capacity and power-supply area of substations in
approximately equal .The network flow model of optimizing
objective year which have been optimized, and quickly find
problem of substation is explained physically in
the optimizing planning scheme in middle years through
computer automatically searching big scopes in a electronical
external
map. tlux
Based on the theory of network flow, a new multi period
optimal selecting algorithm of substation is proposed. The
principle is to attribute the fees of all substations to each arc-
line from the substation to the load point, which sharply reduce
the calculation and avoid possibly non-convergence when
adopting network flow.
A The Mathematical and Network Flow Models of the
Substation Location and Capacity Planning Problems.
fig 2 The network tlow model of optimizing problem of substation
The fee of a substation includes two parts: one is the fixed,
another is the alterable. The former relates to design, using
The right nodes express the load points ,and the external
ground area and iron loss independent of the flux through the
flux at every point is negative , which indicate the needed
component. The other includes the part investment and copper
load ,. These lines represent the distribution lines, and have
loss of the substation .It increases with the growth of the flux ,
the linear fee and the corresponding capacity. Every arc-lines
while the bound fee descends a little . The fixed fee in fee
off the points represent the standby substations which satisfy
function is zero for the established substations. So the
load demand of the optimizing objective year, the value of hF
optimizing problem of the substation in single period can be
and hV respectively represent the fixed and variable fees of
represented as following:
N.SF "\U
present period investing substations. The capacity of arc-line
min Z= c ( h & xx, +hV, x A ) + x ( h V kx f , ) + . ~ C Wxd,/ represents economical supply of standby substations.
!=I k=l ,=I /El,
B. The optimizing problem of substation in the objective year
s.l. C w , < ~ , x e ( ~ , ) x c o s y ,( i = 1 , 2 , . . . ~ ) The optimizing problem of substation in the objective
/ e 1,

f, S X , X C XCOSY,
~ (k=I,2,...NSF) year is described as that: the location and capacity of
f, -< C, x COS^ ( k = 1,2;.. NSU) substation rest on the minimum fee of the investment and
running. We suppose that there exists a load point in the center
f, 2 0 ( k = 1,2,...N)
of every district, and the load is equal to load density
x, = 0 or 1 ( k = 1,2,...NSF) multiplying the area of this district, so the fee of all investment
in equation and running in whole year is shown in equation (1 ).
NSU-numbers of substations utilization c=c,+c, (1)
NSF-numbers of substations future In equation c-
N= NSU+ NSF
f; - t h e load magnitude of substation I, fee of construction investment and running.
c k -the
x, 7 ,
capacity of the No. k substation,
is equal to 1,if substation i is built during the C, = cc
N
Wfd,, - -the estimated fee of lower-
,=I I d #
current stage, or x, is equal to 0,
w ,--the load magnitude in load point j , voltage-network wastage ;
Equation ( 2 - 1 ) is reduced as following equation:

- a87 -
N N are the pending constructing substations and lines. The
rninC=Cfz(S,)+rCCW,d,/ (2) ’
,=I !=I ,El,
calculational result is shown in Tab.11.
s.t. CW,~ s , e ( s , ) c o s p i=1,2,...,N
I€!,

e(&) in equation (2)--the utilization efficiency of the


substation
Therefore, substation location and capacity determine is
abstractively viewed as a combination problem. Adopting
ploughing around and combination optimization can solve this
problem. Meanwhile, every given combination is also a
particular problem. To solve the substation location problem,
the multi-midposition technique is applied to site selection.
Fig 3 Initial Network for distributionplanning
Computer automatically searches the possible location of
substation.
TABLE 11 PLANNINGRESULT FORSUBSTATION
(in MVA)
C. The optimizing problem of substation in the middle years Substations I Period 1 I Period2 I Period 3 1 Period 4
The idea of network flow algorithm to determine the s1 I 15.6 I 22.4 1 22.4 1 22.4
substation location and capacity in middle planning stage is s2 1 21.3 I 21.3 1 21.3 I 21.3
that: the existing and standby substations can be regarded as
linear price. The price of existing substations is hV, and the s4 ’ I 0 0 I 22.5 I 22.5
price of the standby is hR. We attribute the linear price of
every substation to the arc line from the substation to the load The calculational result has shown that this algorithm is
point. The fee of every arc line possesses two parts: the linear high-level efficient and suitable for multi-period extending
price of substation and the fee of the line from the substation planning problems of real large-scale distribution network.
to the load point. The linear pricefi( of standby substation is
as‘sumed, so the price will need to adjust, if it doesn’t accord VI. CONCLUSIONS
with the optimization conditions. The objective function is as
This paper has entirely described the distribution system SLF
following:
and the optimal planning of substation location and capacity.
minZ=CCw,(d,,xr+hR,) (3) The method of the distribution system SLF is brought forward
s.t. C w , <S,e(S,)cosq i=1,2;..,N based upon GIS platform. This method can provide the
I€./, forecasting the load increment of the space, of the time and of
f, S X , XC, X C O S ~ ( k =I,2;..NSF) the quantity in the distribution system. On this condition,
f, S C X~C O S ~ (k=1,2;..NSU) based on the theory of network flow, a new multi-period
f,2 0 (k=1,2;..N) optimal selecting algorithm of substation is proposed, which
could determine the reasonable location, capacity and time that
x, = 0 or 1 ( k = 1,2;..NSF) put into running of the substation. The result indicates that this
From the equations and inequations, we know that the method could preferably meet the need of distribution system
shortest path need not be looked for; the only task is to simply planning.
range the arc line and find the substation and load point from
ranged arc line. If the load of some substation is less than its VII. REFERENCES
capacity and the load point connected with the substation isn’t [ I ] H.L.Willis, J.E.D.Northcote-Green. “Comparison Tests of Fourteen
supplied power for, add the load of the substation again, and Distribution Load Forecasting Methods” [J]. IEEE Transaction on
mark the corresponding load point. When all arc lines have Power Apparatus and Systems. VoLPAS-103, No.6, pp. 1190-1 197, June
1984,.
been found, and check the following inequation: [2] Mo-yuen Chow. H.Tram. “Methodology of Urban Re-development
abs(x x h& - (ht;;+ h y x A ) ) < eps Consideration in Spatial Load Forecasting” [J]. lEEE Transaction On
PowerSysfems, Vo1.12, No.?. pp.996-1001, May 1997,.
If the above inequation is satisfied, the calculation is end.
[3] S.K.Khator, L.C.Leung. “Power Distribution Planning: A Review of
Otherwise, the assumed pricefl( and the arc line fee should not Models and Issues[J]”. lEEE Transactions on Power Systems, Vol. 12,
be modulated and calculate again until the inequation is. No.3, pp.1 151-1 159. August 1997,.
[4] Dai Hongwei, Yu Yixin, Huang Chunhua, et al. “Optimal Planning of
Distribution Substation Locationsand Sizes-Model and Algorithm”.
v. REAL INSTANCE AND RESULTS
International Journal of Electrical Power and Energy System, Vol .IS,
The program is written according to the above algorithm. In No.6, 1996.
[SI H.L.Willis, H.Tram, M V.Engel, et al. “Selecting and Applying
a city of China, Fig.1 is an example of planning distribution Distribution Optimization Methods” [J]. IEEE Computer Applications
network (including four substations, fifty nodes, fifty-nine in Power, pp. 12- 17. January 1996.
lines, four periods), where S1, S2, S3, S4 are substations, the [6] Jiaju Qiu and Jun Li. “Distribution Network CIS”. Automation of
ElectricPowerSystems Vol. 21. No.3,pp. 13-16, 1997.
real line are the existing substations and lines, the broken line

- 888 -
[7] E -C Yeh, H Tram “Information Integration in Computerized [SI V H Quintana, H K Temraz, K W Hipel “Two-stage Power System
Distribution System Planning” [J] lEEE Transacfronon Power Systems, Distribution Planning Algorithm”[J] IEE Proceedins-C, Vol40, NO I ,
Vol 12,No 3, pp 1008-1013, May 1997 pp 17-29, January I993

VIII. BIOGRAPHIES

Zhou Quan was bani in Zigong city of Sun Cnixin was born in Chongqing, China, on
SiChuan province, China, on January 30, 1973 December 13, 1944. He graduated in electrical
He graduated in electrical engineering from engineering from Chongqing University. He
Chongqing University with B E in 1994 and works in department of electrical engineering of
M E i n 1998, and has been assiduously Chongqing University from 1966 to present. He
studying Ph D degrees of Electrical Engineering is a standing director of China Electrotechnical
in Chongqing University, China, Meanwhile he Society and one of directors of Chinese Society
become the faculty of.department of electrical for Electrical Engineering. His current research
engineering of Chongqing University His interests include online detection of insulation
special fields of interest included power system condition and insulation fault diagnosis for HV
planning, DMS, SLF, and GIS apparatus, discharge mechanism in complicated
environment, geography infonnation system and high voltage technique
applied in biomedicine. Hle is an author ofover 200 publication and invention.

- 889 -

You might also like