Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

2006 IEEE PES Transmission and Distribution Conference and Exposition Latin America, Venezuela 1

Optimal Location of Power Transmission


Lines Towers Using Reformulated Dynamic
Programming
F. R. Viera, and J. Toledo H., Member, IEEE

constrains that depends on 3 towers such wind span, weight


span and weight-span to wind-span ratio.
formulated in [1] is reformulated. Results based on this
reformulation and corresponding algorithm and computer II. DYNAMIC PROGRAMMING APPROACH
program are presented. Dynamic Programming consist in
segmenting the route profile, defining sites, and finding optimal
Nemhauser [5] states an alternate exposition of Bellman’s
transmission line from start of profile to each site progressively. The Principle of Optimality “We may say that a problem with
Stages are the sites and the optimal policy will be the line of N decision variables can be transformed into N sub problems,
minimal cost thru each heights of tower that support conductor. each containing only one decision variable”. More
Reformulation refers to redefine optimal policy to represent understandable for us, if we have survey profile adequately
adequately constrains determined by 3 towers (wind span, load prepared for a transmission line. In that profile, a fine partition
span, distance to mass).
is defined of sites where’s possible locate towers. The
problem of to evaluate all combinations of N sites and in each
Index Terms-- ---Dynamic programming. Optimization methods, site its possible locate K heights of towers, with a
Power transmission, Power transmission lines, Transmission dimensionality of N^K is transformed in N sub-problems each
lines, UHV Transmission lines. one can be resolved with one decision variable with K
outages. Problem is run in N stages, i.e., N sites, each one
I. INTRODUCTION with K heights of tower and from each site it’s possible to

G ENERAL standards for UHV power transmission lines


design established in Venezuela recommends computer
software in selecting optimal solution to tower spotting.
reach M other sites determined for the scope of a tower. The
dimensionality is reduced an N x K x M.
As tradition of Dynamic Programming, N represents
Given survey terrain profile on line route and a set of tower stages, K represents states, M sites times K heights reachable
heights and suspension or angle, is required to specify the form a site are decisions and the sequence of states that link a
sites where tower will be erected, the heights of the towers site with the first site and state are called policy. Optimal
and a indication of the class of tower from a set previously policy is the succession of optimal decisions that link the
defined, in such a way that the overall cost of running the line initial tower to the final tower in the last site of profile.
form one end of the profile to the other is a minimum. In Fig. 1 schematically this approach is showed. In site xj
Dynamic programming in [1] is applied to optimal location there are two states, each corresponding to tower height. Each
of the towers in a transmission line design. Other methods state link to other state belong to an anterior site, this link
such computer reproductions of hand design also are in papers represents optimal decision for this estate xj-h1 or xj-h2. In
[2], [3]. Bellman’s The Principle of Optimality tells. “An order to take optimal decision for one state, it must review
optimal policy has the property that whatever the initial state anterior stages and corresponding states, saving for each state
and initial decision are, the remaining decisions must a label that contains anterior stage and state related to optimal
constitute an optimal policy with regard to the state resulting decision. Always those links complain two towers constrains
from the first decision”[4]. Basic algorithm is Dynamic like as ground clearance, distance to objects between state in
Programming and is considered as a finite connected net. examination and optimal decision. Also that link must
A. Constrains. complain three tower constrains like wind span, weight span
and maximal insulator chain deviation, this constrains are
When designing a transmission line spotting, 3 types of verified on the state representing optimal decision. In Figure
constrains may be identified. This are constrains that depends 1, when take a decision in xj-h1, that’s links to x2, the tower
of one tower or related to site. Constrains that depends on 2 in x2 must verify 3 tower constrains, with towers in x1, x2,
towers, such ground clearance and distance to objects, and xj.

1-4244-0288-3/06/$20.00 ©2006 IEEE


2

Fig. 1. Schema of locating towers in Dynamic Programming

III. FIRST POINT PROBLEM


We examine now site-height states such as posed in [1],
and are represented in Fig. 1. Each couple site-height links to Fig. 2 Case test of first point problem
only an optimal decision. Initial site x1, represents a
minimum cost presents in all solutions. When a decision is IV. DYNAMIC PROGRAMMING REFORMULATED
take for height h2, site x1 is selected (supposed that all In [6] is showed a best solution to Dynamic Programming
constrains are verified), it is due that any other decision approach. Best solution is generated in more detailed
should include additional cost of a tower in x2 or x3. When consideration of 3 towers constrains. The quid of question is
this decision is selected, the tower in xj-h2 can be that site-height representation not suffice to represent
compromised in wind load or weight load major part. Win constrains defined on 3 towers.
load compromised can impede to participate in a decision for We reformulate Dynamic Programming approach defining
sites posterior to xj. states as a triplet site-height-range. The range represents the
We can construct a case where exists a prohibited area to preceding sites reachable from couple site xj-hk.
locate towers in sites following to xj, and Dynamic In Fig. 3 triplet site–height-range is showed, the site xj, the
Programming approach as defined in [1] fails in find a height h1, and links representing optimal decisions to ranges
solution. In Fig. 2 show an extreme case where fails Dynamic r1, r2, and r3. Corresponding will have a set of decisions for
Programming approach. In this case, we have 100 meters each height h1, h2, h3…
wind span capacity towers. After meter 100 station, there is a
140-meter area where it’s not possible to locate towers. When
Dynamic Programming approach in [1] is applied on this
profile, in station 100 meters a tower is located that links to
first point. When try to resolve site-height in station 140
meters, can’t to take a decision since a tower in station 100
meters with predecessor in station at 0 meter and successor in
station at 240 meter exceeds wind span 100 meter tower
capacity. The Fig. 2 graphic was produced with DobleRed
(NetDouble) software, which is based in an algorithm that
represents states as site-height-range and finds the only Fig. 3 Schema of solution in Dynamic Programming
solution. To have range included in state definition, permits a reformulated
tower in 100 meter station, to have a state linked to station 0
meter and other link to station 40 meter. When take a V. SOFTWARE DOBLERED (NETDOUBLE)
decision at station 240 meters, find a tower at station 100 By year 2000 is completed in his actual form software
meter with a link to station 40 meter. Wind span on this tower DobleRed. This software implant an algorithm based in
at 100 meter equals (240-40)/2 = 100 meters, which Dynamic Programming model defined in states site-height-
corresponds to tower wind span capacity. range. DobleRed is formed for blocks initial, reading of a site,
selection of optimal arrays for site-height, backward process
to find links for optimal policy.
Algorithm is realized with several versions on solution
catenary’s equation (1), where p is catenary’s parameter. This
is used in determine catenary’s vertex for various load
conditions and for ground clearance.
3

y - yo = p * cosh [ (x - xo) / p ] (1) TABLE I


RESULTS OF COMPARE 9 KM OF TRANSMISSION LINE 115 KV

A. Characteristic of software DobleRed

In a transmission line design DobleRed represents


elements:
1.Station, terrain elevation at line axis, two lateral
elevations, angle.
2.32 heights of towers.
3.32 different class of tower
4.Load conditions for hot conductor, cold and ice
5.Calculations on angle towers.
6.Set of parameters variable for each segment of the line
determined for angle towers.
7.Graphics interface for data preparation.
8.Graphics printed as construction drawings. In Fig. 4 is showed distribution of spans for both cases. For
9.Graphic of calculation progress. DobleRed case, spans are concentrated near 360 meters, while
in other software spans are near 400 meters.

B. Disc space requirement


Number of calculations, memory required and disc memory
required depends of square number of sites dot height number.
In disc space this represents 100 Kbytes for each site
calculated. In real cases limit of 32 bits address is reached,
i.e., 2 Gigabytes for a section of 50 Kilometers, interpolating
one site each 4 meters, with 32 heights of tower.

VI. SOFTWARE APPLICATION RESULTS


Below, results from DobleRed tested with a section of line
to be constructed. A 9-kilometer transmission line at 115 kV
products a low cost of 2.55%, different heights of tower and
mean suspension height lowest. Second case compares a
230kV line S/E Cabruta to S/E Gorrin double circuit and Fig. 4. Test on 115 kV, 9 km transmission line
obtains a low cost for DobleRed 2.23% and a spans B. Test of results on 230 kV double circuit 153 km
distribution more concentrated. After, we show considerations
on sensitiveness of software results to density of sites to locate In a section of 49 km was tested DobleRed with results of
towers. other software. DobleRed produces a transmission line design
with a 2.27% minor cost. In this test, profile of terrain was
A. Compare results over 115 kV 9 km transmission line completely flat, in great extension floodable, this mean that
DobleRed produces 2.55% minor cost over 9 km de survey results are in “design span”. In Fig. 5, a graphic with spans
compared with other software. distribution for both cases. More noticeable in chart is spans
Accordingly to Table I, we observe an interesting fact such concentration in “design span”. Software DobleRed produces
a conductor suspension mean height dismiss, more short spans 73 spans closed to “design span” while other software
and more towers. One of ideas that appoint this case is produces 50 spans.
redesign towers and newly locate the towers, for an important
minor cost of transmission line.
4

reformulation and its application results are presented.


In first term, software DobleRed produce a cut of towers
cost that in all cases was superior to 2%.
Results obtained with DobleRed show characteristics of
high regularity localization such as major concentration of
spans on a value generally called “design span”.
Software DobleRed produce consistent results when test
sensitiveness to density of location sites, best cost and
presents major precision in distribution of spans.

VIII. REFERENCES
[1] [1] G. Mitra, K. Wolfenden. “A Computer Technique for Optimizing the
Sites and Heights of Transmission Line Towers – A Dynamic
Programming Approach”, The Computer Journal, 1968, Vol. 10, p 347 –
Fig. 5. Spans distribution on 49 km of flat terrain 351, Nº 4.
[2] [2] V. Converti, E. J. Hyland, D. E. Tickle, “Optimized Transmission
C. Results sensitiveness to sites number Tower Spotting on Digital Computer”, AIEE Transactions On Power
Software external parameter more important to DobleRed Apparatus and Systems, April 1962, Nº 59, p 55-63.
[3] [3] R. E. Popp, C. J. Dabekis, F. M. Fullerton, ”Electronic computer
is site number to be calculated. It is showed as interpolated program permits optimized spotting of electric transmission towers”
sites in survey profile. That is defined in data as “interpolation AIEE Transactions on Power Apparatus and Systems, June 1963
distance”. In Table II results are showed of 3 runs with [4] [4] R. Bellman, S. Dreyfus, “Applied Dynamic Programming”, book
Princeton University Press, 1962, p 15.
interpolation distances of 3, 4 and 6 meters. When interpolate [5] [5] G. Nemhauser, “Introduction to Dynamic programming,“, book John
a site each 3 meters, cost is lowered in 0.35% compared with Wiley and Sons, 1966.
interpolation distance of 6 meters. [6] [6] F. Viera, M. Pérez., “Localización Optimizada de Soportes de
Líneas de Transmisión de Energía Eléctrica Utilizando Programación
Dinámica”, Trabajo especial de grado. Universidad Central de
TABLEII
Venezuela, Escuela de Computación, Junio 1982
SENSITIVENESS TO SITE DENSITY

IX. BIOGRAPHIES
Francisco Viera was born in Caracas, Venezuela in
1947. Gets degree in computer science in
Universidad Central de Venezuela, 1982, has
worked in computer technical section in Engineering
Dept. of CVG Electrificación del Caroní C. A., Guri
Project in Caroní River development. He worked in
PDVSA, Lagoven, S.A.., actually is independent
software developer.
The same results are showed in Fig. 6 where looks major
concentration of spans near “design span” directly in
proportion to density of sites of localization. Juan Toledo H, (S´89-M´01) was born in Caracas,
DF, Venezuela, in 1961. Electrical Engineer in
Universidad Simón Bolívar USB en 1992. He obtain
degree Magíster Scientiarum MSc, Power systems
option in Universidad Central de Venezuela UCV, in
2k. From 1991 works in CVG. EDELCA, in
Construction Division of hydroelectric Project
Caruachi of Caroní River development. Is member of
several institutions: CIV, AVIEM, IEEE-PES, IEEE-
Magnetics, AsoVac, CIGRE, and SVMNI.

Fig. 6. Sensitiveness to site density

VII. CONCLUSIONS
In this document is presented a reformulation of Dynamic
Programming approach to location of towers in transmission
lines. This reformulation permits to solve the first point
problem, throughout detailed solution to 3 towers constrains.
Additionally, the software DobleRed, is showed based on this

You might also like