Вы находитесь на странице: 1из 7

9th International Conference on Probabilistic Methods Applied to Power Systems KTH, Stockholm, Sweden - June 11-15, 2006

MAINTENANCE PRIORITIZATION OF EXISTING TRANSMISSION LINES USING PRIORITY RISK INDICES (PRI) Ibrahim Hathout1

Abstract: This paper introduces a new model and the associated computer program for the prioritization existing overhead transmission lines for maintenance or refurbishment. The model calculates the priority Risk Index (PRI), which is the product of failure probability, line condition function, and failure consequences function. The failure probability of a line is calculated using the maximum capacities of the line components and the actual weather data from the nearest weather station. The line condition is a fuzzy function obtained using visual inspection and maintenance history of the line. The failure consequences are a complex fuzzy function of many parameters such as public safety, load supplied, impact on customers and system etc. This model would allow asset managers to optimize the use of available resources to achieve the maximum overall return on investment by allocating funds for the lines with highest Priority Risk Indexes (PRI). Illustrative example is presented.
Index Terms - Existing transmission lines, prioritization, failure probability, fuzzy logic, condition assessment, failure consequences, risk indices.

critical transmission lines is not an option, particularly after the 2003 blackout in Ontario and northeastern USA. In the era of deregulation, optimizing resources for upgrading and maintaining transmission lines is of critical importance. The customers of Hydro One expect the power to be there every time they switch on a light. Any failure to meet this expectation can affect the productivity of industries and adversely affect Ontario's economy. However, Hydro One as all other utilities is limited in its means, as it is required to manage operations and maintenance costs.

INTRODUCTION:
The power industry has been going through a major transition over the last decade due to the move towards competition, liberalization, deregulation and privatization. With an increased pressure for bottom line results, utilities and asset owners are turning their attention to return on assets (ROA) and asset management. System safety and uptime remain of prime importance. Transmission lines are the arteries of the power industry. Their health is true gauge of the health of the industry. Utilities understand that failure of

not over-servicing assets that doesn't need it. To establish an efficient repair and maintenance program for the existing transmission lines under financial constrains, it is essential to, first, prioritize these lines to ensure that the finite resources are used to achieve maximum benefit to the grid system and maximum return on investment. Prioritization of existing lines is, however, a very difficult task due to the lack of accurate information, the complex nature of line deterioration, and the uncertainty about the consequences of line failure. Accordingly, prioritization is heavily dependant on the subjective judgement an intuition of senior engineers. Because the procedure of assigning rating information for line components condition and consequences of line failure, the resulting prioritization for lines with similar condition can vary. Consequently, a procedure that can incorporate the subjective judgement (cognitive uncertainty) inherent in line condition assessment and consequences of failures would be useful.

Hydro One uses a proactive strategy to maintain existing transmission assets. This is done by monitoring actual condition through regular inspection, engineering assessment, repair, and coating programs. The goal is to move proactively to service existing assets before it results in costly failure, while at the same time

1 Senior Design Specialist, Hydro One Networks Inc., 483 Bay Street, Toronto, Ontario, Canada, M5G 2P5, email: ibrahim .hathout@Hyd roOne.com

C Copyright KTH 2006

9th International Conference on Probabilistic Methods Applied to Power Systems KTH, Stockholm, Sweden - June 11-15, 2006

Cognitive uncertainty associated with the assessment of line condition and failure consequences can best be handled using fuzzy logic. Fuzzy logic was specifically developed to deal with the fuzziness of human concepts such as those embodied in human perception and decision making. In addition, fuzzy logic provides a systematic framework for dealing with linguistic quantifiers such as "severe, poor, moderate, rusty, fair, very, many, few, good, bad, etc." which arises during assessment line condition and failure consequences.
PRIORITY RISK INDIX (PRI):
To prioritize existing transmission lines for maintenance, it is not enough to know their current probabilities of failures. Asset managers would also like to know the consequences and cost of failures. The failure in this study is the sustained failures from resulting mechanical/structural failures and is not momentarily failures that cause very short interruptions. Although momentarily interruptions are very important to sensitive industries, it is treated under power quality.
A model and the associated computer program are developed to calculate the Priority Risk Index (PRI). The PRI is defined as follows: PRI (i) = Pf(i) x LC(i) x FC(i)

most fundamental problem for design engineers assessing the safety of existing-transmission lines is relating the design safety factor to the failure probability. The safety factor, SF, iS defined as: SF = Rmin/Qmax= (ctD tR)/ (Y VtQ) = (0I1IY) SFC (2)

Where Rmin, Qmax are the resistance and load in the deterministic design, [tRand [tQ are the mean values of R and Q, (D and y are constant design parameters, and SFC is the central factor of safety. If R and Q are independent random variables, then failure probability, Pf, is given by:
Z- z

^,z

Pf = Fzs (-

1 Vz

Where Fzs is the cumulative distribution function (cdf) of the standardized safety margin Zs = (Ztz)/cIz, and Vz is the coefficient of variation of the safety margin. The solution of this equation is given in detail by Hathout [2004].
LINE CONDITION:

(3)

(1)

Where, = is the probability of failure of the i-th line Pf(i) LC(i) = is the condition function of the i-th line FC(i) = is the failure consequences function of the i-th line
It should be noted that the probability of failure, Pf, is calculated for the as new condition and the line deterioration, LC, is introduced as a separate function. The reason is that if the maintenance work is deferred for few years, we need only to upgrade the line condition LC. The probability of failure is not likely to change. FAILURE PROBABILITY:

Conditions of the towers and wires are the two main components used for condition assessment of transmission lines. Inspectors visually assessing the damage condition of transmission towers (foot and helicopter patrol and using the GIS system) often do so using a rating between 1 and 5 where the numbers represent the following: 3 - fair 1 - Very good 2 - good 4 - Poor 5 - very poor These numbers are fuzzy because condition rating would not be interpreted as precisely 2 but rather 'about 2' or poor. These fuzzy numbers are modelled using the membership functions shown in Fig. 1.
V.Good
0

Good

Fair

Poor

V.Poor

0.8
----

X l

0.5
E

It is generally recognized that safety of a transmission line must be statistically evaluated, and reliability or probability of failure has been proposed as a "reasonable" criterion for transmission line safety. Existing transmission lines are designed using experience based deterministic methods with safety factors. The

~0.3 E0 \
0.1

0.4

3 4 Condition Rating

Fig. 1: Fuzzy Condition Rating Functions

C Copyright KTH 2006

9th International Conference on Probabilistic Methods Applied to Power Systems KTH, Stockholm, Sweden - June 11-15, 2006

Since the purpose of this preliminary inspection is line prioritization, detailed inspection, which include climbing towers, measurement of remaining thickness of critical members, hardware and insulation inspection is not needed. The purpose is to have an overview of the structural condition using the following symptoms as a guide:
Very good: Good: Predominant color is grey. Predominant color is grey with some yellow brown stains. Predominant color is greyishyellow brown with some edge corrosion. Predominant color is brown red with corrosion spreading from edges inward. Predominant color is dark-red and corrosion covers almost the entire tower.

COSEQUENCES OF LINE FAILURE:


Sustained interruption caused by structural failures could last from few hours to few days. Engineers do everything they can to avoid structural failures, yet failures will continue to occur due the complex nature of line deteriorations. There are many consequences for line failure that need to be addressed in the risk assessment of existing lines. These are:
Public Safety: When structural failure occurs, the first concern is public safety. In urban locations, utility crews usually reach failure site within 30-minutes to secure the site and retrieve the wires. Since some transmission lines travel through heavily populated urban areas and others travel through unpopulated areas, the safety concern associated with line failure can vary. Following is a proposed guide for safety concern rating - Fig. 2:

Fair:
Poor:
Very poor:

For long lines, it likely that towers condition would vary along the line. In this case, inspector would divide the line into line sections and rate the towers in each section from 1 to 5 (very good to very poor).

Very high:

Samples of the conductors and ground wire are normally taken and tested for tensile strength and ductility (twist or turns).
Similar to towers, wires are also rated from very good to very poor using the following proposed guide:
V. good: Good Fair Poor V. Poor:

Moderate:

High:
Low:

Very low:

Urban areas (near schools, crossing parks, roads, and highways, etc.) Industrial areas Lightly populated areas (small towns) Rural areas and farm land where public exposure is very low unpopulated areas

If NOT> 10 If NOT> 10 If 7<NO<10 If 5<NOT<7 If NOT<5

& TS .100% of BS & 95% <TS<100% of BS & 85% <TS<95% of BS & 75% <TS<85% of BS & TS% <75% of BS

The membership functions in Fig. 2 were selected as trapezoidal not triangular to account for the higher level of uncertainty associated for rating consequences of line failure.

t I 09

V. Low

Low

Moderate

High V. High

Where,>
NOT is the number of turns in the ductile test, TSD0. is the tensile strength, and BS is the rated F0 breaking strength of the wire. 5 0.6 The reason for modelling the wire condition using fuzzy functions is that the sample taken may not be from the worst location. For instance, fair condition vary from poor to good (2 to 4) allowing the inclusion wider range of possibilities. The overall condition of the line can then be obtained using fuzzy weighted average formula.
/\4
-

t\

2 3 4 Consequence of Failure Rating

Fig. 2 Fuzzy Failure Consequence Rating Other parameters that should be considered as a consequence of failures are:

C Copyright KTH 2006

9th International Conference on Probabilistic Methods Applied to Power Systems KTH, Stockholm, Sweden - June 11-15, 2006

Customer load Supplied:


The load rating depends on whether the the load in MW is residential (RL) or industrial (IL). Following is a suggested rating:

Very high:

High:
Moderate: Low:

Very high: High: Moderate: Low: Very low:

IL > 100 MW or RL > 250 MW 50<IL<100 or 150<RL<250 MW 25<IL<50 or 75<RL<150 MW 10<1L<25 or 25<RL<75 MW IL<10 or RL<25 MW

Very low:

2 circuits 500 kV line with circuits utilization > 50% 1 circuit 500 kV or 2 circuits 230 kV with circuit utilization > 50% 500 kV, single or double circuits with circuit utilization < 50% 230 kV, single or double circuits with circuit utilization < 50% 115 kV circuits

Other failure consequences can also be included.


The global failure consequence function is obtained combining all failure consequences using a fuzzy weighted average model. FUZZY WEIGHTED AVERAGES MODEL: To combine fuzzy line condition ratings (wires and towers) and the failure consequences ratings, a fuzzy weighted average model is proposed. The model is based on a fuzzy computational technique called resolution identity of fuzzy sets. The procedure is based on the decomposition of fuzzy sets into non-fuzzy level-sets or intervals by slicing a fuzzy set A at different a-levels or membership levels. An a-level set, denoted by Aa consists of all elements that have a membership grade in the fuzzy set A greater than or equal to the value of a. The fuzzy set A can be written as a function of its level sets as follows:
A = E oc Au (4)

Customer Impact:
Hydro One has two type of customer supply, Network (115 kV to 500 kV) and Radial (115 kV). All major urban centers are Network Supply (NS), while remote customers are normally Radial Supply (RS). If a transmission line in Network supply fails, customers will continue to be supplied through other circuits. In Radial supply, however, if the line fails, outage will occur. Accordingly, failure of radial lines has direct impact on customers, while failure of a line in Network may not have immediate impact on customers but has indirect impact on reliability of power supply. Following is the proposed rating for customer impact:

Very high:

High:
Moderate:
Low:
Very low:

RS heavy industrial use and difficult access RS heavy residential use and difficult access RS moderate use and easy access or 500 kV NS 230 kV NS with easy access or 115 kV NS with difficult access 115 kV NS with easy access

The proposed fuzzy rating model utilizes the simplification introduced by Tee and Bowman [1991] that involves combining the lower boundaries and upper boundaries at different acuts or membership levels.

System Impact:
In Network supply, critical circuits are normally 50% or less utilized. This allows taking some circuits out of service for maintenance or in case of catastrophic failure without interrupting customer supply. However, operators may load some critical circuits above 50% when the demand for power is very high such as very cold days in winter or very hot days in summer. In these cases line failure could have sever impact on the system. Following IS a proposed line rating for system impact:

The overall condition rating of a line is calculated by combining the ratings of structures and the wires using the following fuzzy weighted averages formula:

C-

C ZWi
I

N Wi Where
I

(5)

~ ~ ~(Fig.

an W, is th.soitdrlaiefzywih 3). The relative weight function recognizes that the combined elements are not equal in value, for instance, tangent towers are normally

c.is the condition rating of the i-th element (scture, conductor, or grond wire)

C Copyright KTH 2006

9th International Conference on Probabilistic Methods Applied to Power Systems KTH, Stockholm, Sweden - June 11-15, 2006

designed to fail before conductor. Accordingly, the relative weight for wires should be higher than that for the tangent structure because failure of one tangent tower is likely to be limited to that tower or at most will affect the two adjacent towers if the wires are still intact. If a wire fail, however, it is likely to cause cascading that is likely continue until it reaches a dead end tower. For example if the towers and wires conditions are "poor", then the relative weight for the towers and the wires could be "heavy" and

The program is currently written in Visual C++ and will be converted into Java programming language to allow the program to run under any window platform. The program calculates the probability of failure of the line, the overall line condition, and the global failure consequences function. Then defuzzify the line condition function and global failure consequences function using the Euclidian distance technique, Hathout [1994]. The Priority Risk Index is then calculated using equation 1.

"very heavy" respectively.


V.iight
Light

M odeFrate

Heavy

V. Heavy

.c0.5--E 0.3 0.2

*i0.76
~0.4
0.1 0

1. 0.9 0.8

--

- -

-I

----

Fuzzy Weight

Fig.

4: TLPRI System

Similar formula is used to combine all the failure consequences ratings.


n

Fc Fec=

Where fci is the i-th failure consequence parameter (safety, customer load, etc.) and W, is the associated relative fuzzy weight (Fig. 3). The relative weight functions are very important and should be assigned by senior engineer. It varies from line to line. For instance, if a line travel through heavily populated urban area, then relative weight for safety should be heavier than any other parameter. If the line however travels through unpopulated or lightly populated area, then other parameters may be assigned heavier weight. TLPRI System
TLPRI (Iransmission Lines Priority Risk Indices) is a fuzzy logic computer program and advisor for the calculation of PRI for existing transmission lines, Fig. 4.

___6_ ZWI 1

Wi fci
(6)

The program is designed as a quasi-expert system. The knowledge is stored and can be edited in a separate file ("knowledge File") without changing the system. The user can add more parameters or edit existing parameters. For example in determining the line condition, we suggested that the conditions of the towers and wires be used for the calculation of the line condition. If the user would like to add the condition of the foundations and Hardware he can do so by editing the "knowledge File". The program will search the file and interact with the user to get more information about the ratings and relative weights of all parameters in the knowledge file and calculate the line condition accordingly. The program has also built-in advisor to provide the user with suggestions or hits about the selection of ratings, as shown in Fig. 5. Further, the program gives the user the choice between triangular and trapezoidal membership rating functions. More membership shapes can be added.
Numerical Example: PRI of 115 kV Transmission Line
a

double circuit,

Thel7.6 km line was built in 1922. The total

number of towers is 68 all steel lattice, 1920 type towers. The conductor is 605 kcmil (54/7) ACSR

C Copyright KTH 2006

9th International Conference on Probabilistic Methods Applied to Power Systems KTH, Stockholm, Sweden - June 11-15, 2006

and ground wire 5/16" GS. The wind spans varies from 633' (193m) to 1031' (314m). Visual inspection has indicated that the predominant color of the steel is brown red with edge corrosion. Accordingly the structures were rated poor". Samples of the wires were tested and the results indicated that the wires are also in "poor" condition.

..................

File Optioni

Help

6:... What Is the conductor rating? F.ig 1920.n...... Fi towe...rs Please ec a ratin from Wery good to vey poo Vim Grph i 192 type t

AIf;;sefik
0

utput
int!
00%.1 DT > 10 turns and 90% < TT T 9<
D T > 1 0 turns and T T > = 1

I Very good: G |ryood: Fair: Poor: Very poor:

. Fig.7. 7< DT < 10 turns and 80% < TT 90%. 5 DT < 7 turns and 75% < TT < 80% BS DT < 5 turns and TIT 75% of Breaking Strength (B] iS)

The gust capacities (mph) vs. spans (ft) of the 1920S type towers were calculated and shown in
H

l l
l

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~1

Embr.

f hip iog

h:

C Traezil

r triangular
E

Legend:

Trapezoidal

Triangular

Result

MS

11111111600

650

700

750

800

850

900

950

1000

1050

Span (ft)

Fig. 7: Gust capacity vs. span for 1920S tower


nearest weather station to the line is that at 1 2 3 4 ~~~~~~~~~~The Toronto The annual Airport. average) Fig. 5: TLPRI provide adiseon advise onratingselectionextreme at Fig.5: LPRprvid atig speeds (3-second gust International Toronto International Airport between 1953 and 1983 werearter60.16-mph and reported by (96or.8 Krishnasamy cocrThelie is rtd"ih. On cici the Airoroto .i5H andn8.1 tavl hroghreltiel poultedara, Tabatabai of kmh) the gust data [1988]. Analysis between 1953 and 1983 indicated that the mean andcrosg many passingt close tors hsesFig. and (see Th safet standard deviations of the gust speeds at stretsa parks cig.u6)

slecionthe

Thesine travel tohroughselatiel propulate marea

mhe (13.03 kmh respecivl.8 Ftr ht supply the largest steel producer in Ontario, the annual 1303kextreme gust speeds follower, other circuit is 60 Hz supply TS and DS. The follow load supplied is rated "moderate", and customernternational gust beeds approximately Gumbel Distribution (with interruption is rated "high" since the supply to the ie=0.1589 and f=56.53). This gives a 50-year DS through the 60 Hz circuit must be maintained gust of 65.97-mph (123.49 km/h). The recorded all the time. The system impact is rated "low"

gust speeds at the Airport were converted to a 2 since there are other double circuits 230 kV airor ~ ~ load~ ~ ~ ~~~~~~~~sbrs linele ande (mod0ra4e openme the line location using the following gust speed at loaio) lines on the same corridor, o(1 000/33)s , transfer function Tf = (Ch/1000)n DS through the 60 Hz circult must beflataterrain) where Ch iS the average conductor height, Ol, t2 are the terrain factors (1l=0.21 residential

*~~~~~~~~~

C,gutopyrightmp (2.9kmh.The2006de

9th International Conference on Probabilistic Methods Applied to Power Systems KTH, Stockholm, Sweden - June 11-15, 2006

Results: Using the TLPRI program, the results may be summarized as follows: Failure probability = 0.056473 Line condition = 4 or "poor" Consequence of Failure = 3 or "Moderate"
PRI = 0.056473 x 4 x 3 = 0.677676

Tee, A.B. and Bowman (1991), M.D. "Bridge Condition Assessment Using Fuzzy Weighted V8 Averages", Civil Engineering Systems;n,, March, pp. 49-57.

CONCLUSIONS:
In the era of deregulation, optimizing resources for upgrading and maintaining transmission lines is of critical importance. The proposed model and associated computer program calculates Priority Risk Index (PRI) for existing transmission lines, which is a function not only of the probability of failure and line condition but also of the consequences of line failure. The model and associated computer program would allow asset managers to optimize available resources to achieve the maximum overall return on investment by allocating funds for the lines with highest Priority Risk Indexes (PRI). It should be noted that the transmission line in the numerical example is now being refurbished and upgraded (all wires, insulation, hardware are replaced and the structures are refurbished).

REFERENCES:
Hathout, Ibrahim (2004), "Damage Assessment and Soft Reliability Evaluation of Existing Transmission Lines", proceedings of the 8 International Symposium on Probability Methods Applied to Power Systems (PMAPS 2004), September 13-16, 2004, Ames, Iowa, USA.

HATHOUT: received a B.Sc. degree in Civil Engineering from Cairo University (Distinction with honor), and M.A.Sc. and Ph.D. in Civil degrees engineering from the Universities of Windsor and Waterloo receptively. He is currently a senior design specialist at Hydro One transmission Engineering. Dr. Hathout has extensive experience in design, maintenance, rehabilitation and reliability analysis of structures. His research interests include reliability assessment and applications of fuzzy logic and expert systems to damage assessment of existing structures. He has published over 33 technical papers in the general area of structural engineering and wrote two chapters in two reference books. Dr. Hathout is the recipient of many prestigious scholarships and awards and he is serving in several IEEE committees and working groups.
1k

~~~IBRAHIM

Hathout, Ibrahim (1994), "Treatment of uncertainty in a fuzzy logic expert system for damage assessment of transmission structures", A chapter in the reference book "Uncertainty and Modeling Analysis: Theory and Applications", edited by B.M. Ayyub and M.M. Gupta, Machine Intelligence and Pattern Recognition, Elsevier Science Publishers B.V., North-Holland. Krishnasamy, et al. (1988),"Database for Weather Related Loads on Overhead Transmission Lines in Ontario", Ontario Hydro Research Report # 88-220-K, 1988.

C Copyright KTH 2006

Вам также может понравиться