IEEE Transactions on Power Delivery, Vol. 14, No.

2, April 1999

549

LIMITATIONS OF THE RULING SPAN METHOD FOR OVERHEAD LINE CONDUCTORS AT HIGH OPERATING TEMPERATURES.
Report of the IEEE Task Force "Bare Conductor Sag at High Temperature". Members of the JEEE Task Force of the WG "Thermal Aspects of Overhead Conductors", of the Towers, Poles and Conductors Subcommittee are: Y. Motlis (Chairman), J.S. Barrett, G.A.Davidson, D.A. Douglas, P.A. Hall, J.L. Reding, T.O. Seppa, F.R. Thrash, Jr., and H.B. White. Abstract. This report summarises the work by the Task Force
to review the accuracy of the ruling span method for conductors operated at high temperatures. The basics of the ruling span approximation method have been examined. The traditional ruling span approach can be used with little or no error for a typical overhead line crossing a rolling terrain to predict sags in suspension spans for conductor operating temperatures in the range of 5@C to 7OOC. Sensitivity studies were performed using conductors "Lapwing" and "Tern" in order to quantify such ruling span assumptions as the effect of the longitudinalswing of suspension and line post insulators on conductor sags at high temperatures, and the effect of the suspension insulator string length on the equalization of conductor tensions in adjacent spans. Significant errors in estimating the sag at conductor temperature above lOOOC may occur if the tension differences are not taken into consideration in line sections consisting of a series of spans of non-equal lengths, It was confirmed that the ruling span method is the most practical way to string conductors in multi-span line sections.
c t =

I I . DEFINITION OF TERMS.
A = total conductor area
coefficient of elongation

D = conductor sag D , = ruling span sag

= modulus of elasticity

H , = horizontal tension at initial conductor state

H = horizontal tension Lo = initial conductor length

L = conductor length RS = "local" ruling span S, = ruling span length S, = suspension span length
To = initial conductor temperature

= conductor temperature w = weight of conductor per unit length

HI. THE RULING SPAN METHOD.

Key words: Conductor, ruling span, high temperature, sag,

tension, insulator swing.

A. The basics. The well known parabolic and hyperbolic equations defining
the relationship between span, sag, and tension apply to single level dead-end spans. For a series of spans of unequal length and nearly level elevations, a simple method is needed to determine a theoretical level span length for which the sag and tension characteristics can be applied to determine the sag and tension behaviour of all spans. The solution of this problem was published in 1924 by E.S. Thayer, an electrical engineer in Seattle [11. The solution is now called the Ruling Span method. A common definition of ruling span is a level dead-end span that gives the same change in tension from changes in loading, creep, and/or temperature as that in a series of suspension spans between two dead-end structures [2]. This span "rules" the conductor's sag and tension behaviour for the line section. The ruling span method permits correct sagging of conductors and provides prediction of conductor behaviour with creep, loads and temperatures within the usual operating ranges of 5oOC to 7OOC. The tension variations due to load or temperature changes

I. OBJECTIVE AND SCOPE.

The objective of this paper is to describe the widely accepted "ruling span" method of sag-tension calculation for multiple suspension spans between dead-end structures where the spans are nearly level but unequal of length. Errors due to operation of the conductor at high temperatures and due to imperfect tension equalization at supports is studied and several calculation corrections are noted. In this paper, the high temperature operation means conductor temperature above IOOOC (212F).
PE-I 97-MD-0-12-1997 A paper recommended and approved by the IEEE Transmission and Distribution Committee of the IEEE Power Engineering Society for publication in the IEEE Transactions on Power Delivery. Manuscript submitted July 16, 1997; made available for printing December 12, 1997.

0885-8977/99/$10.00 0 1997 IEEE

550

will depend on the lengths of the spans in the section, and the section as a whole will react to load and temperature changes in the same way as a single "ruling" span [3]. It is a fictitious span with a rate of slack equal to the average rate of slack of the line section. The equation for the ruling span length (S,) of a line section of (Si>suspension spans is:

and is referred in this paper as the traditional ruling span

equation.
The parabolic approximation for the ruling span sag @) is:

longitudinal movement of the attachment point) nor true suspension points (allowing unrestrained longitudinal movement of attachment points). Full tension equalization is unlikely at such points even for small longitudinal movements. 0 at suspension structures supporting a large weight span, tension equalization may not occur even for modest longitudinal movement of the insulator attachment point. for post insulators, tension equalization depends on the combined flexibility of the suspension hardware (if any), the insulator, its attachment to the pole, and the pole. response of strain or suspension structures to varying loads. This can be significant, for example, when tubular steel structures or davit type arms are used.

C. The accuracy.
The ruling span approximation method may not be accurate enough to analyze the operation of a line, although it was used for the design of the line. This is especially true if there is a need to operate the line above the original design temperature. Sag errors caused by incomplete tension equalization between suspension spans result in inaccurate calculations using Eq.1 and Eq.2. This is the main scope of this paper. Sag errors caused by temperature variation along the line section generally cause lesser errors than those due to incomplete tension equalization, and it is outside of the scope of this Task Force. Errors in sag estimates caused by the present methods of modeling of conductor's sag vs temperature relationship (which also affects the sags in individual dead-end spans) may be a future task of this Task Force. While the Task Force certainly does not advocate discarding the existing methods of sag-tension calculations, it has identified situations where line engineers should be aware of the limitations of traditional calculation methods.

The sag-tension behaviour of each of the spans in a line section is determined in the following manner: 0 sag-tension calculations are made for a single dead-end span with length equal to the ruling span (Eq. 1 and Eq.2). the tension in all of the suspension spans of the line section is assumed to be the same and equal to that of the ruling span under all loading and conductor temperature conditions. 0 once the sag @ , ) of the ruling span has been calculated, the sag @) of any other span (S) is calculated as:
D = ( -s )

= D,

(3)

B. The assumDtions.
The ruling span method is called an approximate method because of a number of unwritten assumptions made such as: span lengths are large compared to the difference in elevation of supports. the load per unit length is equal for all suspension spans in the line section. conductor temperature is the same along the line section. the suspension points between adjacent spans are free to move longitudinally without restraint. This is the fundamental assumption of the traditional ruling span method. When circumstances prevent or unduly restrict this free movement and tension equalization, sag predictions based on the ruling span method may be inaccurate. Other errors resulting from the ruling span approach may be caused by: angle suspension insulators, running angle insulators and inverted "V" strings are neither true strain points (allowing no

IV. SINGLE DEAD-END SPAN.

Sag-tension calculations can be complex even for single spans with fixed end points. The conductors' non-linear elasticity, thermal elongation, plastic creep elongation, and the various combinations of ice, wind, and temperature conditions may need to be considered. Useful information can be obtained from the tensiontemperature relationship in a level dead-end span, considering only the elastic properties and thermal expansion of the composite conductor, and making simplifying approximations.

551

The length ( I , )of conductor in span using the parabolic approximation, by:

(S)can be calculated
(4)

Arc elongation (or slack) is defined as the excess length of conductor ( I , )relative to the span length (S)and is given, in the parabolic approximation, by:
(5)

The change in strain in a single dead-end span, in which the temperature changes from T, to T, accompanied by a change in tension from H, to H, is:

---

&-Lo- H-Ho
Lo

As

+a (T-To)

in the insulator string, the greater the restriction. The ruling span method assumes that complete equalization is achieved, thus overstating the insulator swing. Suspension point movement is usually, but not always, less than that calculated using the ruling span method. Depending on the specific spans lengths, there may be a difference in horizontal tension between any two adjacent uneven spans. The ruling span inaccuracies are largest for lines with short insulator strings [5], since the ruling span approximation assumes an infinite string length. The shorter the insulator's length, the greater is the restriction on movement. High operating temperatures (over l W C ) further degrades the accuracy of this approximation. A complete analysis of multi-span line sections should take into account conductor properties, spans, line profile, line angles, insulator string properties, support stiffness, original sagging and clampingin procedures, weather loading history, creep, and a reasonably good knowledge of the existing condition of the line section under study.

The "graphic method" of sag-tension calculation [4], assumes that the change of slack is euual to the change of strain. Using this assumption we can combine Eq.5 and Eq.6 to obtain Eq.7:

VI. NUMERICAL EXAMPLES.

The basic assumption for the numerical examples is that the initial position of the insulators is vertical, either without or after offset clipping.

f i
This cubic equation in H describes the approximate tensiontemperature relationship for a single dead-end span. If H, and Consider a dead-ended line section consisting of 10 spans, conductor "Lapwing", ACSR, 4517, total area = 1590 kcmil (1.249 in2 or 805.8 mm2); weight=1.792 lblft (26.2 N/m), RTS=43780 lb (194.7 kN), E,=9.5*1@ psi (65.5*103 MPa), a= 11.6*10-6 1PF (21*10-6 lPC), E,,=27.5*106 psi (189.6*1@ MPa). Initial condition: ?,=8440 lb (37.5 kN) @lO"C; S,= lo00 ft (305 m); final condition: Td, = l W C . Suspension insulator string is 5 ft (1.52 m) long; and its weight is 120 lb (534 N). Since the six computer programs used by the Task Force to calculate sag, tension, and swing at high temperatures showed very close results, Table 1, Appendix I, lists the average values of those calculations. As can be seen, the sag in the longest span of 1500 ft (457.2 m) is 4.9 ft (1.5 m) smaller when the tension differences are taken into account. In the span of 1150 ft (350.5 m), the sag @lOoOC is 1.4 ft (0.4 m) larger when the tension differences are considered. The explanation of these results is very important for line engineers and is given below. It has been noted that short and long spans react differently to changes of temperature. Short spans are more sensitive to temperature changes than long spans. For the 1150 ft (350.5 m) span in our example, the positive sag error (when the

T, are known e.g., measured in field, the horizontal tension H may be computed for any given temperature.

V. MULTI-SPAN LINE SECTIONS.

Real overhead lines are not limited to spans with fixed endpoint supports. In a typical transmission line, most spans are "suspension" whose end-point supports move, coupling each span with adjacent spans. Transmission lines are usually sagged to maintain the insulators plumb. With temperature rise, creep and permanent strain from weather loads, the conductor elongates. When a line section has spans of differing lengths, the conductor elongation causes the insulators to depart from their vertical position. As temperature increases, the suspension point moves toward long spans and away from short spans to equalize horizontal tension. Figure 1, Appendix I, shows an insulator string and the forces acting at the suspension point. In general, at a suspension point between two spans, the movement of the suspension point caused by tension difference is restrained by the vertical load at the suspension point. The larger the load

552

actual sag is larger than the sag calculated using the ruling span method) depends on the span's length, the tension differences in adjacent spans, and insulator string lengths. Only for the idealized ruling span method, stiffness is independent of insulator string length. The tensions in the 1150 ft (350.5 m) span are modeled by the ruling span method to follow those of the lo00 ft (304.8 m) ruling span, but the restraints of the insulator strings cause the span to behave more like an 912 ft (278 m) span. The resulting behaviour is described later in this paper as a "local" ruling span which differs from the traditional ruling span method.

string length, in the same spans, there is practically a complete tension equalization. This confirms the significant effect of the length of the insulators.

MI, OTHER APPROACHES TO RULING SPAN ERRORS.

High temperature sags can be modeled with alternative techniques such as "local" ruling span and fits to tensiontemperature behaviour. A better understanding of these techniques may be beneficial for the users of the real time line monitoring systems.

Case 2. Effect of line Dost insulators deflection.

Analysis using the case described below shows that the current practice can lead to sag errors, because although flexible, a polymer line post insulator is still a magnitude stiffer than a suspension insulator for a similar span. Calculations were made using conductor "Tern", 795 kcmil (403 mm?, ACSR, RTS=22100 lb (98.3 kN); w=0.8958 lb/ft (13.1 N/m); diameter = 1.063 in (27 mm); total area=0.6674 in2 (430.6 mm?. The ruling span = 500 ft (152.4 m). Initial tension is 15% RTS, i.e., 3315 lb (14.75 kN) @lOOC. Line post insulator is 4.36 ft (1.33 m) long, and has a stiffness of approximately 2500 lb/ft (3728 kg/m). Table 2, Appendix I, shows the effect of deflection of line post insulators on sag at high temperatures. The sag errors with the 2500 lb/ft line post insulators are approximately 1/2 of the difference between the ruling span and individual span cases. Stiffer line post insulators would cause the line section behave more like individual spans. For comparison, seven units insulator strings in the same span would have spring constant of about 120-150 lb/ft (1750-2190 N/m). The sag errors would be much larger if the spans lengths were increased.

A. "Local"ruling man. The concept of a "local" ruling span ( R S ) is to find a deadend span that has the same tension-temperature relationship as each actual span. In order to define such a local ruling span, rewrite Eq.7:

Case 3. Effect of the insulator string length.

The basic assumption of the traditional ruling span method was verified i.e., as the insulator string length increases to infinity, the tension @lOSC approaches the ruling span tension. Calculationsperformed for Case I assumed an insulator string length of 5 ft (1.5 m). Similar calculations and comparison were made using the insulator string lengths of 2.5 ft (0.8 m), 14 ft (4.3 m), and 200 ft (61 m). The later can be considered as an insulator of infinite length. At conductor temperature of lOoOC, the 5 ft (1.5 m) long insulator string swing results in a tension difference of 182 lb (0.8 kN) between spans of 750 ft (228.6 m) and of 1150 ft (350.5 m). For the same conditions, the swing of the 2.5 ft (0.76 m) long insulator string results in tension difference of 339 lb (1.5 kN). For the assumed 200 ft (61 m) insulator

If (To, H,) and (T, H)are known for a particular span, its "local" ruling span is obtained using this equation. This is a single-parameter fit to the tension-temperature relationship to two known points and is only valid to the accuracy to which the line section can be modeled (e.g. angle structures, elastic response of structures or uncertainties of elastic modulus and coefficient of thermal expansion). H , ) and (H) are determined by measuring Alternatively, if ( the conductor tension directly [6] or derived from the measured sag at two known temperatures, an accurate single parameter fit can be established between the two fitted points. In both cases, the "local" ruling span's tensiontemperature behaviour will differ from the actual ruling span except between and near the two fitted points. If necessary, this second order deviation can be calculated using a multispan program and fitting "local" ruling spans for each span of interest. However, it may not be practical to have ruling spans which vary with temperature and location. The "local" ruling span length was calculated using Eq.8 for Case 1. The calculation was performed for a temperature change from To= 10C to T= 100C, and the corresponding tension change is from &=8440 lb (37.54 kN) to H=5886 Ib (26.18 kN).

553 4. For overhead lines planned for high temperature operation, it is recommended to add a buffer of about 1 m to the vertical clearance at maximum thermal sag. 5. Line post insulators can cause errors if the sag calculations are made using traditional ruling span equations when the span lengths vary significantly. In many cases, post insulators can cause individual spans to behave as if they were dead-ended at every structure. 6. The ruling span concept remains the most practical method to string overhead line conductors. The ruling span effects are not dependent on the type of conductor but rather on the amount of tension change per degree of conductor temperature change.

Because the "local" ruling span had the temperature-tension behaviour of a span shorter than lo00 ft (304.8 m), the sag in the 1150 ft (350.5 m) span was actually larger than that estimated using the traditional ruling span equation,

In general, as a result of the insulator swing at high operating temperatures the spans can interact in such a way that a multispan sag-tension program may be necessary to predict the line section sag-temperature behaviour more accurately. In some cases it may be practical to run a multi-span program once and fit the results. One way to fit the tension-temperature of each span is to rewrite Eq.7 as:
C T=a +bB+-

IX. ACKNOWLEDGMENTS.
Task Force members would like to thank the following persons and organizations for their contributions to the success of this task, namely: D.E.Koonce, C.B. Rawlins, N.P. Schmidt, W.A. Chisholm; for providing the results of calculations using the following computer programs: STRING, Power Technologies Inc.; SAGSEC, Power Line System Inc.; NIP & TUCK, Linesoft Inc; SPRING, The Valley Group; SWING, Ontario Hydro; and RECONSTRUCTION, Dr. L.M. Keselman.

(9)

Si1

This equation is linear in the parameters a,b, and c, which can be fitted by linear regression to either the results from a multispan sag-tension program or to observed values of (T)and (H). The fitting of (a) takes into account the constant terms H , J and (To); the fitting of (b) takes into account involving ( the "springiness" of all the other spans either increasing or decreasing the effective spring constant of the conductor and; the fitting of (c) is similar to fitting a "local" ruling span. If is required as a function of (T),either a closed-form solution of the cubic equation or an iterative solution may be used. Once (H) has been obtained for a given span, the sag may be computed using the actual span lengths and Eq.3.

X. REFERENCES
[ 11. E.S. Thayer, "Computing Tensions in Transmission Lines", "Electrical World", no.2, pp.72-73, 1924. [2]. C.O.Boyse, N.G. Simpson, "The Problem of Conductor Sagging on Overhead Transmission Lines", Journal AIEE, Vol.91,Part 11, pp.219-231, Dec. 1944. [3]. P.F.Winkelman, "Sag-Tension Computations and Field Measurements of Bonneville Power Administration", Journal AIEE, Vo1.78, pp. 1532-1548, February 1960. [4]. Theodor Vamey, ACSR Graphic Method for SagTension Computations (Book), Aluminum Company of America, Pittsburg, Pennsylvania, 1927. [5]. W.A. Chisholm, J.S. Barrett, "Ampacity Studies on 49OC-Rated Transmission Line", IEEE Transactions on Power Delivery, Vo1.4, no.2, pp. 1476-1485, April 1989. [6].T.O.Seppa, "Accurate Ampacity Determination; Temperature-Sag Model for OperationalReal Time Ratings", IEEE Transactions on Power Delivery, Vol.10, No.3, pp. 1460-1470, July 1995.

VIII. CONCLUSIONS.
1. The traditional equations describing the relationship between temperature and span length, sag, and tension are fully valid for dead-end spans only. A multi-span line section can be analyzed as an equivalent single dead-end "ruling" span. 2. The traditional ruling span method can be used with acceptable error margins for lines which are operated below l@C and have relatively equal and near level spans. 3. When old lines, originally rated for low operating temperatures, are uprated for operation at higher temperatures, the magnitude of sag errors should be evaluated using one of the available computer programs. The main source of errors is the longitudinalinsulator swing in line sections with unequal spans.

554

Appendix I.

tan 8 = H / (wi/2+we),
where: 8 = angle of insulator swing
- 7

I
I
I

a \ d
I

(conductor

wi

weight of insulator

wc= total weight of a vertical span of

conductor

'

*.

H = total horizontal load including component of tension due to line angle

I4
Fig. 1. Suspension Insulator Swing

S,=lOOO ft (304.8 m): 700 1150 m 213.4 350.5 Sag@lOOC 18.1 48.8 basedon s, 5.5 14.9 Sag@100"C 19.1 50.2 5.8 15.3 w/swing effect Sag error 1.1 1.4

span,

ft

0.3

0.4

@lo0 "C, sagz36.84 ft (11.23 m), tension=6090 lb (27.1 kN) 750 450 900 750 950 1500 850 228.6 137.2 274.3 228.6 289.6 457.2 259.1 20.7 7.5 29.8 20.7 33.3 83.1 26.6 6.3 6.3 10.1 25.3 2.3 9.1 8.1 8.1 31.7 22.2 21.7 33.4 78.2 26.9 6.8 2.5 6.6 10.2 23.8 8.2 9.7 1.5 1.9 0.6 0.96 0.2 0.3 -4.9 0.5 0.2 0.6 0.3 0.1 -1.5 0.1
~~~~~

650 198.1. 15.6 4.8 16.2 4.9 0.6 0.2

Table II Effect of Post Insulator Deflection on High Temperature Sag.

span,
,

ft m

Sag@lOOC based on S , Sag@ 100C w/deflection Sag@ 100C

individual span Sag error

Sp500 ft (152.4 m): @lo0 "C, sae21.12 ft (6.44 m), tension=1328 lb (5.9 kN) 575 430 530 390 600 5 10 3 90 5 80 350 106.7 175.3 131.0 161.5 118.9 182.9 155.5 118.9 176.8 7.3 19.8 11.1 9.1 21.6 16.8 15.6 9.1 20.2 2.2 6.0 3.4 5.1 2.8 6.6 4.8 2.8 6.2 8.6 19.2 11.6 16.8 9.8 20.1 9.9 15.5 19.4 2.6 5.9 3.6 5.1 3.0 6.1 4.7 3.0 5.9 9.4 18.2 12.3 16.2 10.8 19.3 15.4 10.8 18.4 2.9 5.5 3.7 4.9 3.3 5.9 4.7 3.3 5.6 1.2 -0.6 0.6 -0.09 0.66 -1.5 -0.12 0.7 -0.8 0.4 -0.2 0.2 -0.03 0.20 -0.5 -0.04 0.2 -0.2

400 121.9 9.6 2.9 10.5 3.2

11.1
3.4
0.6

0.2

555

Discussion

B. Freimark (American Electric Power. Columbus, OH): AEP recognizes the errors associated with the traditional Ruling Span method. but we have not found a better procedure, nor does this paper suggest a better procedure, other than to "add a buffer amount of 1 meter (3.3 feet) to the vertical clearance" at the maximum operating temperature.
In the example of a series of unequal (tangent) suspension spans preseiited in Case 1, the general trend is for spans shorter than and approximately equal in length to the Ruling Span to have adjusted sags approximately 37 cm (1.2 feet) more than anticipated, and spans significantly longer to have sags up to 1.5 meters (5 feet) less than anticipated. Regarding the reduction in sag in the longer spans of the line section, in my experience when an overhead line has significant variations in span length (other than when crossing a specifichnique geographic feature such as a major river). the line is traversing very rugged terrain and "jumping" from ridge-to-ridge. For longer spans, the minimum clearance usually does not occur near the mid span (where the valley ''drops" away) but nearer the supporting structures (approximately at the quarter points of the span). Therefor. the larger sag at the mid-span (where the clearance is large) has a minimal effect on the final structure heights. and excess structure height is generally not occurring. Since 1986, we have included the effects of elevated temperature creep in the design of new lines and the review of older lines for operation at elevated temperatures. For the conditions outlined in Case 1, this increases the design sag of the Ruling Span by approximately 30 cm (1 foot). Not mentioned in the paper is the effect of "room temperature" creep due to the time lag between putting the conductor in the stringing blocks and actually sagging and clipping in the conductor. If the conductor's NESC-Heavy design tension were a conservative 40% of its rated strength and the wire was pulled-in and hanging in the stringing blocks at 70% of the sagging tension for 12 hours (overnight). 20% of the 10-year room temperature creep would occur, increasing to 25% after 48 hours and to 30% after 120 hours (Creep being a log function.). For the conditions specified for Case 1 in the paper, the difference between the Initial and Final sags At 16C (60F) is approximately 1.5 meters (5 feet). Therefor: if the creep is ignored (as is normally the case) and the wire is then sagged to "Initial" sadtension values. the "Final" sags would be less than the design values by approximately 30 cm (1 foot). which. again, is generally considercd to be conservative. Additionally, since 1978 at AEP, we have added a 61 cm (2.0 foot) buffer to the NESC required clearances at the conductor maximum operating temperatures (to account for errors in surveying, structure setting, conductor sagging. etc.).

Therefor. AEP has effectively been adding more than the "1 meter" (3.3 foot) buffer recommended in the paper: 30 cm (1 foot) from elevated temperature creep +30 cm (1 foot) from room temperaturG creep before sagging + 61 cm (2 foot) AEP buffer = 1.2 meters (4 feet).

Also not directly stated but only inferred to in Case 3 in the

paper is the relationship of insulator string length to unbalanced tension. As the insulator string becomes longer, the differences in tension in adjacent spans decreases (generally inversely to the increase in the length of the insulator string moment arm). Thus the sags will be closer to the Ruling Span based predictions. Therefor. this issue is of greater concern at lower line voltages and less of a concern at higher Line voltages. Manuscript received February 13, 1998.
R. 0. Kluge (Wisconsin Power and Light Co., Madison, WI 53701): The IEEE Task Force on "Bare Conductor Sag at High Temperature" have presented a very timely and useful paper for utilities desiring to operate their transmission lines at maximum the capacity. By comparing sag calculations using both a ruling span and what the authors call a "multi-span" approach, they demonstrate the limitations of the ruling span method. They also explain that this limitation is inherent in the definitions of ruling span method, that is, the ruling span method assumes an idealistic support that has no horizontal restraint where as the multi-span method considers partial restraint of the insulators. The authors then show that, if the line is operating at elevated temperatures, the conductor may not have uniform tension because there is horizontal restraint at the insulators.

The reader should be cautioned that the comparisons shown in Table I and I1 contain the premise that the conductors had uniform tension between dead ends, at the time of installation. Since this premise may not always be true on a real line, the sag differences could be greater or smaller than shown in these tables. To account for this, the authors adeptly explain that a "second order deviation can be calculated using a multi-span program and fitting 'local' ruling spans for each span." They add, however, that this "may not be practical."
As impractical as it may seem, at times, it is necessary. From my experience analyzing the sag of actual lines, there are numerous events that can cause the conductor tension to be different between individual spans even when at normal ambient temperatures. These include:
1) The conductor may not have uniform tension when clamped at the time of installation.

2) Settlement of structures or relaxation of anchors or guys.

3) Prior heavy ice loading on the conductors.
4) Line alterations including structure height increases or relocation of structures (unless the conductor is reclamped or sliced to restore the balance tension condition).

556

5 ) Numerous methods of increasing the conductor clearances

such as off-set clipping at critically low spans or removing a segment of conductor from a critically low span. Any of these conditions may result in either an increase or decrease in conductor tension for a portion of a line that could result in unbalance tension. A sufficient number of spans should be measured to obtain actual field conditions and knowledge of any unbalanced tension. Of course, if unbalance tensions exist, multi-span methods as recommended by the authors are even more necessary. Item 3) presents special problems because the utility may not know the ice loading history for the conductor. The magnitude of a heavy ice loading e\ ent is important because it may cause additional inelastic strain in the conductor that exceeds the long term creep. The obvious result of this strain is increased sag. However, since, at icing temperatures, the strain will generally be greater in the aluminum, the steel will thenceforth support a greater share of the conductor weight. At high operating temperatures, this is significant because steel has a smaller coefficient of thermal expansion than aluminum and, therefore, the change in conductor sag with temperature will be less. In other words, after a heavy ice loading event, the ACSR conductor will behave more like SSAC (steel supported aluminum conductor).

M. J. Tunstall (The National Grid Company, UK): This paper provides a useful commentary on the ruling span approximation and draws attention to a problem that is not well-recognised among many practising line designers. With the active uprating of lines taking place in many parts of the world, driven by the increasing need to make the maximum use of existing right-of-way, the paper is both timely and germane.
In one or two places, however, the explanation of the role of suspension insulators seems a little confusing. The fundamental assumption of the ruling span (also known as the equivalent span) approximation for a multi-span section is stated in Section I11 A: that the tension in all of the spans is equal and changes by the same amount, as temperature or loading changes. In practice, this is only true either if all the spans are identical or if the suspension insulators are infinitely long. In the latter case, the insulators are required to swing only an infinitesimally small amount to achieve tension equalisation across the insulators. It is self-evident from Figure 1 that a finite length insulator can sustain a swing only if there is a tension difference between the adjacent spans, thereby conflicting with the ruling span assumption. The ruling span approximation is therefore likely to be accurate only where both the true solution requires very small insulator swings and the suspension points are free to allow them to take place. Subsequent statements, such as in Section V: The ruling span method assumes that complete equalisation is achieved, thus overstating the insulator swing are therefore confusing. NGC are currently uprating a number of their AAAC circuits from a rated temperature of 75C to 90C. The procedure that has been adopted is to check the clearances of the line at 90C using standard software, based on the ruling span approximation. If this procedure leads to predicted clearances that are within 0.5 m or less of the design value, then the corresponding sections are checked using the NGC program, GenCat, which allows for insulator swing and for the reduction of EA with increased temperature. 0.5 m is typical of the maximum differences that have been found between ruling span and GenCat calculations. Where necessary these checks are combined with surveys of critical spans, during which direct measurements are made of conductor temperature. Manuscript received February 13, 1998.

To determine where the conductor is on its final strain curve, multiple sadtension measurements must be taken at differing temperatures. Tension monitors, as suggested by the authors to measure conductor tensions when operating at elevated temperatures, possess an addition feature, that is, they readily provide multiple tension measurements. This can be extremely valuable not only to calculate the long term conductor strain but also record ice loading data to determine retum intervals of local ice loading events.
Utilities should be aware that many of these field conditions I have mentioned can have a greater effect on the conductor sags than the choice of a numerical method of evaluation. This is not to say, that utilities should not use a more precise method of evaluating conductor sags. Conversely, the experiences I have identified above provide additional reasons for using one of the multi-span programs referenced by the authors. Two final questions: If one does not know the ice loading history, would you comment on the number of additional sag measurements required to back calculate the strain level of the conductor? Since the paper did not provide a comparison of the programs, are all of the programs equally capable of handing imbalanced initial tensions or span specific tension modifications?
1 R. J. Carrington, New Technologies for Transmission Line Upgrading, IEEE-TP&C ESMO Conference paper, March 1998.

2 T. 0. Seppa, et. al., Use of On-Line Tension Monitors for Real Time Ratings, Ice Loads and Other Environmental Effects, CIGRE, 1998. Manuscript received February 13, 1998.

Ronald J. Carrington (ECSI Corporation, Spokane, WA): The authors have done a fine job of helping focus attention on a most important subject. The numerical example presented in Case 1 is a set of unequal spans with the conductor support points all at the same elevation. It is preferable that the authors use a numerical example with the point of support points at different elevations so that the reader could see the effects of differences in elevations as well as unequal span lengths. The authors suggest that all six computer programs showed very close results which leads the reader to believe that any of the six programs are adequate for this type analysis. The software programs

557

identified in the paper have not been rigorously evaluated and representing that they are all produce equivalent results is misleading the reader. A numerical example with conductor support points not all at the same elevation would present a more realistic example of typical lines found in the field. Item 5. in section VIII. CONCLUSIONS, recommends a clearance buffer of 1 m for high temperature operation. This paper has presented insufficient evidence to support this recommendation. The clearance buffer, whether used for design or analysis should take into account how well all of the input parameters and calculation methods are known. Examples could be presented where the appropriate clearance buffer should be less than 1 m and examples could also be presented where the appropriate clearance buffer should be far in excess of 1 m. The clearance buffer should be established for a particular transmission line after careful engineering study. Providing this rule of thumb in this paper is inappropriate, misleading and potentially dangerous. In paragraph C. The Accuracy, the authors state that sag errors caused by temperature variation along the line section are generally less than the sag errors caused by incomplete tension equalization. This is also a very important topic particularly for rerating and real-time rating. Since this statement has been included in this report without documentation or reference, the authors are encouraged to support this claim in this report or a subsequent document.

1m proposed in Conclusion 4. We do not agree that such an IEEE Task Force should be recommending such an important parameter without much more intensive discussion. For a new line planned for high temperature operation, we suggest that RS method be applied to derive sagging data and to spot the structures and that the structure heights then be adjusted by means of one of the H. T. programs or equivalent as mentioned in this paper. Then the sagging in can be done with due consideration given to application of offsets. The needed buffer should then be no more than and probably less than anything used in the past for normal temperature operation. For HT operation of an existing line, the first imperative is to determine the position of the existing wire, clamps and insulator string inclinations and with this knowledge, almost all need for a buffer such as I m has been removed. The step to HT operation will contain some small areas of doubt regarding wire properties but not warranting a buffer of 1m. We look forward to more detailed discussion of this question of necessary and sufficient buffers.

Manuscript received February 23, 1998.

J. Stephen Barrett (Ontario Hydro, Toronto, Canada): Although a member of the Task Force on Bare Conductor Sag at High Temperature, I would like to add a discussion to clarify how to fit Eqn. 9 to the multi-span tension-temperature relationship. The three-parameter fit of Eqn. 9 is best determined by a least-squares fit to all the data points, but a simple fit to three points may be more convenient. Given three points, (To,Ho), (T1,Hi) and (T2,H2),define:

H.BRIAN WHITE, Consultant Hudson, Quebec, Canada :

While being an active member of the Task Force we expressed and maintained some opinions contrary to those of the majority on several issues discussed in the document. It was suggested that we present our views after presentation of the document which we do at this time. One theme runs throughout the document and that is that the Ruling Span (RS) method is somewhat deficient in fulfilling its purpose, a concept mentioned in the abstract and also in item 6 of the conclusions where it states that the RS concept remains the most practical method etc. Except when dealing with an absolutely level series of even spans, the RS method is the only method for deriving necessary sag tension data for spotting the structures, an iterative process (usually one iteration is sufficient) that provides the sag tension data for stringing and obtaining vertical strings . . at clipping in if supplemented by apphation of offsets; the offsets being needed to compensate for the unequal horizontal tensions that exist in the mix of long and short spans or in spans at different elevations while in the stringing blocks. Neglect of offsets is quite rational and common if they are small but their omission will certainly complicate the process of trying to predict the sag distortions that will occur at higher temperatures. Assuming correct installation with offsets so that insulator strings are vertical at clipping in, the entire system begins to distort as soon as creep enters the picture. Our second item of concern relates to the subject of buffers, and ,specifically, the recommended buffer of about

d=T,-T,

e=H,-H,

f =1/H:-1/H:

p = T2 - T, q = H , - H , r = 1I H ; - 1I H: z=er-qf Then b = (dr - p f ) I z c = (ep - q d ) I z a=T,-bH,-clH~

To provide an example of the accuracy of the local ruling span fit of Eqn. 8 and the 3-parameter fit of Eqn. 9, they were applied to the 450 ft. (137.2 m) span of Case 1 in the paper. The first tension column provides the actual tensions, Ha,, in the 450 ft. (137.2 m) span, based on a multi-span program. The second column of tensions, H,,, is based on the ruling span method, where the ruling span is 999.93 ft (304.8 m). The tension error increases significantly with temperature. The third column of tensions, H,,, is based on the local ruling span fit of Eqn. 8. In the papers example computation of the local ruling span, the spring constant AE is given as 1.249.9.5.106. which is the aluminum component alone. The total spring constant, including the steel, is given by 1.249.9.5.106 + 0.086.27.5.106. Based on H, at 10C and 100C from Table I, the local ruling span for the 450 ft. (137.2 m) span is 860.88 ft. (262.4 m). In this case, the local ruling span fit is within 67.5 lb. (300 N) of the actual values.

558

TABLE I: Tensions in the 450 ft. (137.2m) Span of L:ipwing ACSR in Case 1

H,,

1L-J

(W
8440.0 37.541 7999.6 35.582 7600.9 33.809 7240.2 32.204 6913.7 30.752 6617.8 29.436 6349.0 28.240 6104.4 27.152 5881.1 26.159 5676.8 25.250

spans between deadends. The most interesting actual test case appeared in 1994 with an installation at a 100 kV line section consisting of only four spans, 1183 ft., 975 ft., 300 ft. and 1787 ft. The tension variation measured at the 1183 f t . span indicated a local7Ruling Span of 1 160-1 190 ft., compared to the classical Ruling Span of 1399 ft. The multi-span tension program calculated local Ruling Span value of 1180 ft, i.e. in close agreement with the measured values. The outline of this procedure which we call Ruling Span Calibration was presented at IEEE-PES WPM-95 in a report at TPC subcommittee meeting. The importance of the local Ruling Span concept and its experimental determination is that it is a fundamental tool for accurate determination of sags in sag-critical spans of the line section from tensions or sags measured somewhere else in the line section. The formula (8) of the report and a description of its use for real time ratings was first published in [3]. [l] Tapani 0. Seppa: A Practical Approach for Increasing the Thermal Capabilities of Transmission Lines ZEEE Transact. on Power Delivery, Vol. 8, No. 3, July 1993, pp. 1536-1550.
[2] Tapani 0. Seppa: Accurate Ampacity Determination: Temperature - Sag Model for Operational Real Time Ratings IEEE Transact. on Power Delivery, Vol. 10, No. 3, July 1995.

The last column of tensions, H,,, is based on the 3- [3] Tapani 0. Seppa: Power Transmission Line Monitoring parameter fit of Eqn. 9, using the 3-point fit method described System. U.S. Patent 5,5 17,864, May 2 1, 1996. above. Using the values of Ha,, at 10C, 5OoC and 100C Manuscript received February 13, 1998. from Table I, the resulting values of a, b and c are: a = 29.07051, b = -0.008712633 and c = 3.879675.109. In this case, the 3-parameter fit is within 6 lb. (27 N) of the Y. Motlis: actual values. The authors would like to thank Mr. Freimark for his comments regarding buffers. AEPs experience in using Manuscript received February 23, 1998. buffers and with the effect of the insulator string length on sags is consistent with our paper findings and recommendations. We would like to emphasize that the whole T. 0. Seppa (The Valley Group Inc., Ridgefield, CT): issue of sags and tensions in significantly inclined spans like Although a member of the above Task Force, and in complete in high mountains is very different than for level spans and, agreement with the TF document, I would like to add a therefore, is not addressed in our paper. Buffers are normally discussion to clarify the concept of the local Ruling Span. added to predicted sag values to take into account possible computation errors and effects that have not already been The data from tension monitoring systems [l] indicated as considered in the sag calculations. Although creep was not early as 1992 that in some cases the measured tension discussed in the paper, it is normally included in both ruling variation as a function of conductor temperature was not that span and multi-span analysis, and so a buffer for creep is not expected from Ruling Span calculations. Report [2] describes one method of experimental determination of local Ruling usually required. In ACSR conductors, elevated-temperature Span length which is applied for determining the conductor creep is often less than creep at lower temperatures because temperature from the measured tension of a line section. The the aluminum stress decreases with temperature. We would report also points out that there are other factors than the therefore question why a buffer for elevated-temperature insulator swing, such as line angles, structure response, creep has been specified. A buffer for creep before sagging-in deadend insulator properties, elevation differences and the is also not required, firstly because the effect can be computed uncertainty of conductor properties which affect the and secondly, because creep before sagging-in, like prestressing, is actually beneficial in reducing long-term creep calculation. and sag. Normally, a 61 cm (2 ft) buffer is adequate. But in In 1992 we developed the earliest complete multi-span sag cases where is a mixture of short and long spans with short model, which allowed analysis the effect of all the above insulators and the ruling span method is being used, we have factors for tension variation within line sections of up to 24 suggested a 1 m (3.3 ft) buffer.

559

Mi. Kluge has drawn attention to a number of situations that can result in unbalanced tensions between spans. Although these were not dealt with in the paper, he points out that they can be included in multi-span methods. Mr. Kluge also discusses the problems associated with not knowing the conductors loading history, and the benefits of monitoring tensions or clearances. If the conductors loading history is unknown , the permanent strain can be inferred from a few sag and temperature measurements along the stringing section, assuming that the stringing conditions are known (from old line layout drawings, for example) . Although we are not familiar in detail with all six computer programs referenced in our paper, we believe that at least some of the multi-span programs can handle initial tension imbalances. In response to Dr. Tunstalls comments, the authors agree that tension equalization is always imperfect for insulator strings of finite length. This does not, however, render the ruling span approximation less useful since, as Dr. Tunstall notes, field measurements of sag are typically within 0.5 meters of calculated values (though he does not indicate the spans, sags, or tension levels used). The point of the paper is that the assumption of tension equalization through insulator swing is seldom perfect but often adequate. We thank Dr. Tunstall for pointing out an area of possible confusion in our paper. The source of the confusion appears to be that swing has been interpreted to mean swing angle, whereas the authors intended it to mean the distance that the bottom of the insulator string moves. The tension at which equalization occurs at support points is a direct function of the horizontal displacement and not of the swing angle. On another hand, the swing angle is a function of the insulator length and not of the horizontal displacement of the support point. As insulator lengths increase, the swing angles decrease for the same horizontal displacement; or the horizontal displacement increase for the same swing angle. The ruling span method, therefore, overestimates the horizontal displacement. Mr. Carrington raises the qucstion of the effect of elevation differences of structures on the results of the study. We have compared the 10-span test case of level terrain to another situation, in which the first five spans are in a +lo% slope, followed by the last five spans in a -10% slope. In effect, this means that the 1.6 mile (2.6 km) line section would climb over a 400 A (120 m) ridge. The difference in sag errors @lOOC conductor temperature to the values shown in the Table 1 of the report (level span) and non-level spans is from +O. 1 ft to -0.1 ft (3 cm) in all spans but in span no. 5 where the difference is 0.3 R (9 cm). It indicates, though, that in most cases the eKcct of relatively small slopes on tension equalization is much less than the error caused by the longitudinal inclination of suspension insulators. Obviously, in a truly mountainous terrain a multispan sag-tension program is rcquired for accurate calculations. The purpose of the paper, however, was not to promote any or all of the multi-span programs, but to illustrate the limitations of the ruling span method for lines crossing average rolling terrain in accordance with the Terms of References of this Task Force supported by all of its members. Mr. Carrington inquires about the sources of the statement

that the errors caused by tempcrature variation along line section are generally less than sag errors caused by incomplete tension equalization. This is based on several published reports which indicate that the temperature variation within a single span is gcnerally larger than the variation of the average tempcrature of spans in a line section [1,2]. [2] states: We discovercd.. that if a temperature sensor is placed each mile on a 20-mile transmission line, the variations in the measured tcrnpcrature within the span was greater than the variation among the spans. . We are pleascd to note that Mr. Carrington and Mr. White do agree with us that, in gencral, it is essential to specify a clearance buffer. However, thcy are questioning the Task Force recommendation of about 1 111 (3.3 ft) buffer for overhead lines planned for high temperature operation. The written discussions to this papcr by Mr. Freimark and Dr. Tunstall, as well as verbal discussions (Mr. W. Peters) during the presentation of the Task Force Report in Tampa, Florida, February 1998, are in support of that use of buffers is a prevailing practice. A recent survey of 47 North American utilities, conducted in conjunction with a CIGRE survey, indicated that the average clcarance buffer used by these utilities for 115-138 kV lines is -3.2 ft (-1 m), and slightly larger for lines at 230 kV and above... If the recommended in our paper about 1 m buffer is intcrpreted as a rule of thumb, we would like to use thc opportunity to clarify for all readers that it is rather a typical value for design of new lines and for the existing lines that have not had a condition survey. Our reply to Mr. Freimark illustrates the point that the buffer needs not be a fixed value. A careful engineering study mentioned by Mr. Carrington means to us evaluation of the condition survey of an existing line. The above discussions imply that for high temperature operation of existing lines, an accurate survey and subsequent analysis could be used to reduce buffers. This assumption is not generally warranted. For example, data in [4] to Mr. Carringtons discussion describes line uprating to 95C based on sag measurements using laser. The line section is very uniform and in level terrain, with little variation in the span lengths. Even then, the results show a sag error range of 1.4 f t (0.43 m) at 95OC, depending on which of the measured data and what calculation method is used. Mr. White is suggesting including offset clipping as a part of this report. The Task Force members have agreed that offset clipping is a very involving topic by itself and, therefore, should not be a part of this task. We agree with Mr. White that the initial position of the suspension clamp is very important to perform longitudinal swing analysis at high operating temperatures. To address this issue, the following sentence is included in the paper: The basic assumption for the numerical examples is that the initial position of the insulators is vertical, either without or after offset clipping. That is, during the design stage, if the suspension clamp is deflecting along the line, it should be brought back to the vertical position using offset clipping, then, to perform swing analysis. Prior to the emerged issue of sags at high temperature operation, many experienced line engineers were adding a vertical clearance buffer to compensate for well known inaccuracies in survey, design, drafting, tower

560

erection, structural flexibility, sag/tension calculation method, conductor stringing etc. Additional problems surfaced with operating line conductors at high temperatures.. We do not agree with Mr. Whites statement that because of the application of offsets ..the needed buffer should then be no more and probably less than anything used in the past for normal temperature operation... When designing a new line, due to the above mentioned inaccuracies or combinations of them, a clearance buffer should be added no matter offset clipping is applied or not. Mr. Carringtons statement that providing in our paper this rule of thumb, e.g. 1 m buffer, .. is inappropriate, misleading and potentially dangerous. . could be addressed to the Mr. Whites statement that knowing the position of the existing wire, clamps and insulator string inclinations results in that .. almost all need for a buffer such as 1 m has been removed... We would like to re-assure Mr. White that the ruling span method will be used as the most practical and the only method to design and string new lines. Dr. Barretts comparison of the local ruling span and the 3 parameter fit methods should be helpful to readers who arc trying to decide bctwccn thc two methods. It does occur to the authors that the diffcrcncc between the two methods is much smaller than the differcnce bctween actual sags and those found by either approach. It also seems to the authors that sag errors at conductor tcrnpcratures (e.g. 5OOC) well below the maximum (lOOCO are of limited interest to line designers since thcy do not impact the vertical clearance at high temperature opcration. Mr. Seppa is to be thanked for providing additional comments on the concept of local ruling span. As pointed out in his references, it is possible to improve the accuracy of high temperature sag calculation by calculating the tension variation between suspension spans. It should be pointed out, however, that the cases whcrcin this is necessary are limited to those where the ratio of suspension span lengths is quite large. In the example that Mr. Scppa cites, the ratio of the

longest to the shortcst suspcnsion span is nearly 6. This is certainly an unusual line dcsign wherein the use of the d i n g span approximation quite rcasonably lcads to large errors in calculated sags. The local ruling span concept is useful for the evaluation of data logged by a real tinic line monitoring system. However, it should be seen in a propcr pcrspcctive. It is not fbndamental in the sense that it can be computed from the span lengths in a line section. Thc starting point is to make the approximation that the changc of arc elongation in a single dead-end span is equal to the change of strain, as expressed in Eq. 7 of our paper. This is the basis of the Graphic Method by T. Varncy, dcveloped in 1927. The formula appears in Winkclmans paper of 1960 and has been used in various sag-tension programs. The local ruling span formula, Eq. 8 of our papcr, is simply a rearrangement of the same old formula, Eq. 7. Thc local ruling span concept is to use this formula to fit a span to the tension-temperature relationship, which is either computed by a multi-span program or observed on an actual line. In the example in our paper, the long span of intcrest has the tension-temperature relationship of a short span, lcading to a short fitted local ruling span. So, the short local ruling span does not explain the unexpected bchavior; it is the unexpected behavior that explains why thc local ruling span is short. It should be pointed out that most utilities with lines in mountainous terrain have had complete multi-span sagtension programs for dccadcs. References. [ 11 T.O. Scppa, A practical approach for increasing the thermal capabilities of transmission lines, E E E Transactions on Power Delivcry, vol. 8, no. 3, July 1993. [Z]M Davis, Discussion rccord , Panel on Dynamic Thermal Ratings, IEEEPES SPM, 1982, San Francisco, Ca.
Manuscript received August 13, 1998.