0142-0615(95)00054-2

Electrical Power & Energy Systems VoI. 18, No. 3, pp. 175 183, 1996 Copyright ~!- 1996 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0142-0615/96/$15.00 + 0.00

ELSEVIER

Application of geographic information system for substation and feeder planning WM

Lin, M T T s a y a n d S W W u

Department of Electrical Engineering, National Sun Yat-Sen University. Kaohsiung, Taiwan. ROC

a single substation location which yield the minimum feeder loss 1 5. Other approaches deal with a wider area with the use of complicated mathematical formulations including optimization algorithms from linear programming, transportation models, branch and bound, to mixed integer or quadratic programming techniques 1-8. However, the physical constraints of the land use or street accessibility were not easy to model and were very often neglected. Thus, the results could be misleading. With a GIS, this paper presents a method with simple mathematics and the use of actual street routes for distribution planning. A growing city was used for the case study. Load distribution maps were used to define load points. A load point is defined by lumping the load between two neighbouring switches and is assigned to the switch closer to the maintransformer. In addition, with the data and system setup, planners are facing a friendly environment where graphic user interface (GUI) was used and the buttons/icons were designed for MMI.

In this paper, an effective method is presented to determine the substation siting and feeder planning based on the shortest path. A geographic information system (GIS) with distribution database, state-of-the-art computer graphics and a proposed method are integrated to solve the sizing and siting problems of distribution substations, and to perform feeder routing, including load transfer to the neighbouring substations. A GIS provides an environment for building a database which contains both graphic information and non-graphic facility data. A distribution system of the Taiwan Power Company (Taipower) in Kaohsiung ci O, was selected for the test. It can be shown that the proposed process will pro vide a physical en vironmen t and a friendly man-machine ir,:terface ( M M I ) for planners to perform the distribution planning. Keywords: distribution system planning, geographic information system, mimmal-path algorithm, load switch pattern

I1. A M / F M / G I S I. Introduction

system overview

An AM/FM/GIS is an integrated hardware and software system which is used to create and manipulate digital maps and related data. With the support of the relational database management system (RDBMS), the attribute data of electrical facilities, such as feeder capacity and installation date, can be retrieved, modified, and displayed with graphics. The application of GIS for performing the automated mapping (AM) and facility management (FM) has been implemented by many utilities9-11. Besides basic AM/FM functions, several distribution application functions, such as work-order management, trouble call dispatching system for outages, and the substation plan, etc. have been developed to support the automation of system planning and operation 9-11. Figures la and lb show the concept of GIS and how the GIS was used by the authors to support the engineering applications. A data preparation program is required between the GIS and the applications to retrieve the

The load demand in the Taipower System has increased substantially during the last decade owing to the rapid economic growth on the island. The determination of substation locations and[ capacities for future expansion are crucial from the viewpoint of effective long-term planning of power distribution systems. Often a substation may face a capacity shortage due to the load growth. The way to solve the problem is to add new substations or perform the load transfer to the neighbouring substations. By doing so, the territory served by all the feeders of the substation could vary due to the load redistribution, i.e., the excessive loads in a substation service area may be transferred to the neighbouring substations or a new substation. A number of methods concentrated on the planning of Received 25 March 1993; accepted 5 April 1995

175

176

Appfication of GIS for substation planning. W. M. Lin et al.

Composite of - Street map - Land use map - Hydro map - Utility network map

External Programs

(b)

~

~ Outputs

Data ] Preparation

RetrieVeData l

III. The proposed model for distribution system planning Figure 2 shows the flowchart of the proposed process. First, a GIS environment has to be set up including the hardware and software. Both the graphic data and text data have then to be prepared. Graphic data are in the form of maps including street maps with tax parcels, city planning maps, power network facility maps and load distribution maps. The former two could be obtained from the city while the latter two are utility maps which were digitized for this project. Text data associated with the above maps were also imported, keyed, or calculated as described in Section IV. With load forecast and current service territory of each substation as known information, the optimization was processed in the following main steps:

(a)

Proper Data Format

data and to update the GIS database. Through GUI of GIS, the output results are easily displayable on graphic terminals.

~ Update

I GIS Database [ 4 _ _ ~ Graphic User lnterface(GUl) GtS Environment Figure 1. (a) GIS system. (b) GIS integration

(1) (2) (3) (4) (5)

overall system power balance and violations check; substation sizing and siting; optimization of substation service area; feeder routing optimization; cost estimation.

The constraints used for system violation check are as follows: Mainstransformer Maximum allowable capacity 25 MVA Maximum allowable feeder numbers 5 Substation Maximum allowable capacity 25, 50 or 75 MVA Number ofmaintransformer 1, 2 or 3

Data Preparation and Database Design -Map Digitization -Attributes Construction -Load Forecast

No

Yes • iSubstation Siting and Sizing I Service Area Optimization

¼ ~ Feeder Routing Optimization

~ Cost Estimation Figure 2. Flowchart of computerized procedure

A study area is a district with its existing system, i.e., original service areas, substations and feeder routings. For a horizon year, if the system is short of power, new capacity will be added by either the expansion of existing substations or the addition of new substations. Service areas will be adjusted after the determination of proper substation siting and sizing. Feeder routings will follow the above assignment. The procedures are explained in the following sections with service area explained first.

II1.1 Service area optimization Service areas are divided by the assignment of load points. An area is the territory served by a substation with certain load points. To minimize the loss, load points should be assigned to the nearest available substation. Generally, a straight line was used for calculating the distance between a load point and a substation, which is a straight forward but non-practical way for most planning techniques. With sufficient data and GIS, physical constraints with street accessibility and right-of-way could be modelled, especially in a suburban or city area as studied by this paper. A shortest-path algorithm based upon dynamic programming was used in this paper to determine the distance between a load point and a substation along the city streets. According to the physical distance, a load point is assigned to a specific service area. Figure 3 shows an example which includes two load points A, B, two substation S1, $2, and the potential feeder routes with the right-of-way and accessibility

177

Appfication of GIS for substation planning: W. M. Lin et al. -- -- -

4'

(1) assigned to a single substation; (2) covered by more than one substation; (3) uncovered by any service area because of capacity shortage.

Accessible streets with right-of-way

I

t

Considering the capacity levels, each load point could be:

CityStreets

I I

Optimal feeder routing with shortest distance

q

X

If there are uncovered load points, more capacity will be added according to the levels described above. On the other hand, if a load point is covered by more than one substation, this point will be assigned to the closest substation with the shortest distance calculated above.

Figure 3. Example of practical feeder routing Table 1. The distances of potential routes on a map

For Point A

Distance (m)

For Point B

Distance (m)

A-s-B-v-S2 A-s-b-v-x-S2 A-s-v-x-S2 A-S-v-S2 A-u-S 1 A-u-t-S1

20 23 20 23 17 18

B-s-A-u-S1 B-s-A-u-t-S1 B-s-u-S1 B-s-u-t-S1 B-v-S2 B-v-x-S2

28 29 29 28 9 12

111.2 Substation siting For substation siting, the load centre will be found and compared with the nearest potential sites of Taipower. A method ]5 was used to locate the load centre which is the optimal site with minimal feeder losses. The site (Xs, Ys) of the load centre is calculated as follows: n

Xs-

~50MVA

n

~

/i/

1

O

•

o~•

Ei=1

(Xs, Ys) coordinates of load centre,

//\

*

o

t - s - i=ln

(Xi, Yi) coordinates of load points i, number of load points in the study area, demand of load point i.

Li

Oo

~

where

Service areas at each level

Neighboring Substation

I*\

i=1n

}2 i=l

Point A is assigned to SI, the shortest distance is 17m. Point B is assigned to $2, the shortest distance is 9 m.

7 5 B ~

n

~ L 2 y.2 Z..a ~ z

L2 j(.2 Z...a i l

t

If the non-served load points are scattered around the whole district, the point with maximum load is chosen for the new substation, and is compared with the nearest potential site of Taipower. Figure 5 shows such a choice.

*/

~Served load point

. on serd,o,,oint

111.3 Optimal feeder routing Once a service area and its load points are decided, the load switch pattern is used to determine the load allocation to feeders with the shortest path. Figure 6 is an example for a load switch pattern algorithm with six load points. Starting from the substation, the set of 'connected nodes' denoted by S contains one node (substation) initially. All other load points are in the set of 'unconnected nodes' denoted by S, i.e.,

Figure 4. Capacity levels and service areas along city streets. Table 1 shows the length of the potential routes between load points and substations. The lengths are calculated with a shortest-path algorithm. It is shown that load points A and B should be separately connected to substations S1 and $2, respectively, with the calculated distances of 9 m and 17m. For the planning of service areas or the assignment of load points, the substation capacity has to be considered. Capacity levels for the substations considered are 25, 50 and 75 MVA with one, two, or three 25 MVA maintransformers. Figure 4 shows the service areas with different capacity levels.

s = {s/s} = {a, B, C, D, E, F} Then, the distance from any node in S to the node in S is computed and the one with the shortest distance is added

10MVA

+

© 1M V A

O

o 3AI

2.SMVA

Non-Served Load Points

I

[]

FI I

I I

I

Potential Substation Siting Maps

Figure 5. Determination of substation site for diversified non-served load

Feasible Substation Site

Application of GIS for substation planning. W. M. Lin et al.

178 S: A

B

~

shortest street : distance B

?

A

\

1

C

D

1

~oo~Xl~

/

E

/

to S. Let C be this node. The sets S and S now become

F

s = {s/s, c}

~q0,~jf~

S = {A,B,D, E,F}

s

a) S and S- set A~ B

D

E

F

shortest street : distance

D F

S:

E

b) S and

S set with C in S

A new feeder connecting load C to the substation S/S is implied in this process. Now, calculate the distance of every node in S to every node in S. For example, node F is the next to be added to set S, and the distance S / S - F is 500 m and C - F is 100 m where Node F should be connected to Node C, i.e., no extra feeder would be required. The feeder S/S-C has to be extended to F. The sets S and are now:

Figure 6. Example for load switch pattern

s = { s / s , c , F} g = {A,B,D,E} This procedure is repeated until all load points are connected, i.e., all the nodes are in S and S = ~ . Figure 7 shows the complete network configuration. Constraints used for the load switch pattern for feeder configurations are as follows:

C Eo//

Figure 7. Complete computation

F

network configuration

Cost(S)

after

• Maximum 300 A flow of each feeder. (300 A is the rated current for 477MCM conductor) • Network radiation: Only one feeder is allowed to connect a toad. • Maximum 5% voltage drop deviation allowed. • Maximum of 5 feeders are allowed for each 25 MVA transformer. • Maximum 25 MVA for each transformer. 6MVA=v/3

CO3 ~ CO2 CO1

[

1

I I I

x ll.4kVx300A

6 MVA is the constraint used for load capacity for each feeder and the voltage drop of each load point is obtained as follows:

I

I

I

25

50

75

MVA

CO1=99278"103 NT$ CO2=(99278+19956)*103 NT$ CO3=(99278+19956+19956)*103 NT$

Figure 8. The step cost of a substation

%AV=[V@omlxLx(RcosO+XsinO) xlO0% where S L R

flow capacity of feeders in MVA, length of feeder between two points (km), resistance of feeder (f~/km),

Potential substation sites of Taipower Figure 9. Street map of the study system

Application of GIS for substation planning. W. M. Lin et al.

"

179

r-"

•

: Substation

• : Load Point

:

Feeder

Figure 10. Original service areas and the network configuration of the study system X reactance of feeder (fl/km), 0 angle of power factor with pf = 0.8, Vnom rated voltage of feeder (kV). Table 2. Description of the study system Substation name

Feeder name

No. of load points

Feeder demand

Growth rate

Neiwei

MA31 MA32 MA33 MA34 MA35

22 25 8 2l 22

7420 5360 1600 4581 4520

0.16 0.13 0.12 0.11 0.15

12 11

Kaohsiung

LG31 LG32 LG33 LG34 LG35

4080 5140 5040 4340 5580

0.15 0.08 0.05 0.15 0.10

2'7 14 :~

111.4 Cost estimation Constructing a new substation is certainly more expensive than expanding an existing one. The cost is considered as a step function as shown in Figure 8. The value in New Taiwanese dollars (NT$) was acquired from Taipower. Costs CO1, CO2, and CO3 are the costs of a substation with one, two or three transformers. The cost for constructing a new feeder depends upon the length of the feeder. For the variable operating cost of a feeder, the loss of power has to be calculated as follows: Ploss = I2RL Pcost = Ploss × UC

Table 3. Sample of the GIS text data record ID

Data name

Data type

Description

Data source

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

TAG Name X Y Type MVAmax MMApre DemandL Distance On/Off R X Ampmax Cost Level Reserve

Character Character Real Real Character Real Real Real Real Integer Real Real Real Real Real Real

A key label Facility name x-coordinate y-coordinate Substation level Max. MVA value Present MVA value Load demand value Distance between two points Switch On/Off Feeder resistance Feeder admittance Feeder current maximum Facility cost Substation capacity level Reserved space

Input Input Generation Generation Input Input Input Calculation Generation Input Input Input Input Input Input

Appfication of GIS for substation planning." W. M. Lin et al.

180

F ' ~ H i g h cost land

]

]Non-available

[pz'~] Low cost land land

Qualified choice

Load center Figure 11. The qualified site near the load centre is chosen for the new substation where Ploss Pcost I R L

UC

with six 3600:1 scale E-size maps used. Figure 10 is a physical layout of the network with street map turned off. Two sites were chosen as potential sites by the long-range plan in the late 1970s. Table 2 shows the name, number of load points, feeder demands, and forecast growth rate of each feeder in the study system. A simple average growth rate was used for illustration.

loss of real power (kWh), cost of losses (NT$), current of feeders (amp), resistance of feeders (~/km), length of feeder (km), unit cost of kWh (NT$).

~ Pi+! - Pi

Total cost, including the variable cost and the fixed costs of the substation and feeders, were calculated during the process.

IV. C a s e s t u d y The system under study is a distribution system in Kaohsiung, containing two substations, ten feeders, and 170 load points. Figures 9 and 10 show this study area

annual growth rate -

where Pi is the historic peak load of year i in kVA and n=10. The demand and installation capacity of the load points were obtained from the load distribution map of Taipower. The calculation is described in Appendix I.

Neiwei Substation I Kaohsiung Substation Service Area I Service Area

New Substation Service Area

$

$

! mm: Substation

e-i n-1

• : Load Point

Figure 12. Service areas and feeder configuration after planning

: Feeder

Application of GIS for substation planning: W. M. Linet al.

( 1 ) Plot l e e d e r I N I - I I In color I 3 I (2) Plot ~eeder I N 1 - 2 1 in color [ 4 I (3) Plot [eeder [ N 1 - 3 l u-t color [ 5 ]

(4) Plot leeder l N! -d, l in color l 6 I (5) Plot l e e d e r [ N l - 5 I in color { 7 I ( 6 ) Plot f e e d e r [ s | - I I in color [ ~ ] }

[

original ss, s t a m

[

[ [ [

s/s expansion s/s slUnal re-dec,dn

] ] ]

1 81

{--;o~-d~t,to. ---I l~ :

[

data output[ ~ - - ]

.

.

.

.

Text o u t p u t

] ]

r~r~r~

1

,

# Name : "Total feeder" # Description: All f e e d e r s in this d i s t r i c t will be # d i s p l a y e d after the e x e c u t i o n of this command.

# # free a > /dev/null map=Qmap #colortab reset #wact w3 #rawp land co i0 er wact w2 el7 #colon wact wl ezc

rawp land co i0 print "$CLS" plot urbanNl plot urbansl plot urbans2 plot subl plot sub2 plot subnewl reset A sample p r o g r a m of the button "Total Feeder" in F i g u r e 13

Figure 13. MMI of the main display

The two existing sub:stations have a 25 MVA maintransformer each and no space for expansion. The new design will adopt 3 × 25MVA standard. It is already clear that it is necessary to construct at least one new substation foJ: the horizon year because of load growth.

IV.1 Text data Table 3 is an example of the text data required for this study. It includes the name, data type, description and data source of each record. The T A G is used to associate a Key attribute with a graphic feature, and can be

accomplished during digitizing. A common record was designed for this study to describe the substation, feeder, and load point. Data could be either generated from graphic data or input from a keyboard as described in Table 3. For example, x and y coordinates and distance, could be generated from the graphic database, and name, MVA and resistance must be keyed in. Load demand has to be calculated by preset equations. Input data acquired from TPC, was updated in 1991. The GIS actually stores the network model which could be used for power system simulation and displays. Data access routines were used to retrieve graphic and

Application of GIS for substation planning. W. M. Lin et al.

182

Table 4. Computational results

Loss (kWh) Substation name

Before planning

After planning

Total cost after planning (NT$ × 103)

Neiwei Kaohsiung New

1089883 1028460 -

638437 616006 566173

13879 13392 122551

Total

2118343

1820616

149822

nongraphic data and data were sorted in a predesigned format for applications. After execution of the application program, the database could be updated with the results ready for viewing through the graphical display. IV.2 Test and results The total load of the system was 48661 kVA in 1991 and the maximum installation capacity of the study system was 50 MVA. For the five-year plan, the total demand will grow to 87008 kVA. A new substation is needed with capacity of more than 38347 kVA, i.e., 50 MVA. Considering the land use, hydrology, soil type, and land cost, a number of sites are suitable for new substations. The substation siting, chosen by the shortest-path algorithm and the load switch pattern, and the service areas and the feeder configuration are shown in Figures 11 and 12. Figure 13 shows the M M I of the main display. A button is a collection of macro commands. The design of a button is also shown here for illustration. Table 4 shows the loss and cost. From Table 4, the total loss after planning was less than that before planning, and the total cost after expansion.

V. Discussion and c o n c l u s i o n This paper describes the use of GIS for distribution planning. An optimization procedure considering physical constraints was developed to determine the sizing and siting of substations and feeder routing based on the shortest path. Data conversion was done for the GIS database which includes graphic data and attribute data. An interface program was developed to retrieve the facility data for the proposed application which runs the shortest path and load switch pattern algorithm. Many other application programs could be integrated with GIS to share such an efficient environment. In addition, a good GIS could also provide tools for M M I customization. It can be seen that with the integration of a GIS, the planners are provided with a new way of looking at the distribution system.

4 Sun, D I, Farris, D R, Cote, P J, Shoults, R R and Chen, M S 'Optimal distribution substations and primary feeder planning via the fixed charge network formulation" IEEE Trans. Vol PAS-101 No 3 (1982) 602--609

5 Thompson, G L and Wall, D L "A branch and bound model for choosing optimal substation locations' IEEE Trans. Vol PAS-100 No 5 (1981) 2683-2687 6 Marshall, A C, Boffey, T B, Green, J R and Hangne, H 'Optimal design of electricity distribution networks' l E E proc. C Vol 138 No 1 (1991) 69-77 7 Adams, R N and Laughton, M A 'Optimal planning of power network using mixed-integer programming' l EE Proc. C Vol 121 No 2 (1974) 138-148

8 Hindi, K S and Brameller, A 'Design of low-voltage distribution networks: a mathematical programming method' l EE Proc. C Vol 124 No 1 (1977) 54-58 9 Dirrwiddle, B R 'AM/FM provides a records management solution, but not without problems' Transm. Distrib. (April 1989) 32-36 10 Pistorese, T 'Installing geographic information system requires careful planning' Transm. Distrib. (October 1990) 60-66 11 Johnson, N A 'MGE sets standards for AM/FM/GIS project' Transm. Distrib. (October 1990) 36-47

Appendix I One feeder of Figure A.1 is used as an example. The installation capacity of load point A is 700 kVA, load point B is 800 kVA, and the peak load of the high voltage customer is 300 kVA. Assuming 1000 kVA is the recorded peak at the substation, the demand of each load point is calculated as follows: demand of load point A AC -AC+BcX(PD-HD) 700 x (1000 - 300) 700 + 800 = 326.7 kVA demand of load point B BC -AC+BcX(PD-HD) 800 × (1000 - 300) 700 + 800

-

= 373.3 kVA

NC

VI. R e f e r e n c e s 1 Crawford, D M and Holt, B S 'A mathematical optimization technique for locating and sizing distribution substation' IEEE Trans. Vol PAS-94 No 2 (1975) 230-235

2 EI-Kady, M A 'Computer-aided planning of distribution substation and primary feeder' IEEE Trans. Vol PAS-103 No 6 (1984) 1183-1189 3 Wall, D L, Thompson, G L and Northcote-Green, J E D 'An optimization model for planning radial distribution network' IEEE Trans. Vol PAS-98 No 3 (1979) 1061-1068

NC

NO

NC

NC

700 KVA 800 KVA 300KVA

~: GQ 1"~

High Voltage Customer

NC : Normal Close

: Load Point

NO : Normal Open

: Subststion

I'1 : Circuit Breaker

Figure A.1. Feeder model when switches are closed

Application of GIS for substation planning: W. M. Linet al.

Thus, AC BC PD HD

rated installation capacity of load point A (kVA); rated installation of load point B (kVA); peak demand of substation (kVA); rated installation capacity of high voltage customer (kVA).

183

The feeder peak load, rated installations and high voltage loads are available from the load distribution map. Load at each bus has to be calculated by the use of the above formula.