Experts say utilities are pushing transmission systems to unsafe limits.
When transmission-line rating assumptions do not match the physical realities of transmission networks, the consequences can be disastrous.
Timelines of line trips in Ohio and Indiana preceding the Aug. 14, 2003, blackout indicate that many lines failed because of phase-to-ground faults. Such faults can occur only if the lines are operated well above their design temperatures-due to inadequate clearances to trees and other objects under the line-or most likely, because of a combination of both conditions.
A news release of Aug. 22, 2003, from American Electric Power (AEP) reveals important information. It shows that at least four, and more likely, six transmission lines had phase-to-ground faults between 3:41 and 3:51 p.m., while the line loadings were substantially below assumed safe ratings. A later article in the identifies a total of 64 breaker actions in Ohio between noon and 4 p.m. on Aug. 14.
We believe that AEP bases its rating assumptions on actual ambient temperature, full sun, and a 3 ft./second wind speed. Compared with those assumptions, ambient temperatures in the area were slightly lower than assumed, but the sky was generally clear. PJM uses ambient-adjusted emergency ratings, with a wind speed of 1.5 knots (2.5 ft./second), while some utilities in the Midwest use wind speed assumptions up to 4.4 ft./second. Thus, an important question: Were these assumptions valid at the time of the emergency loads?
Traditionally, the engineering and planning functions of the transmission owners have determined thermal line rating assumptions. Unfortunately, these persons often have little or no understanding of meteorology, and they generally base their rating assumptions on weather data from National Weather Service sites. A quick review of weather data from National Weather Atlas shows that the median summer wind speed in Ohio averages about 12 ft./second and that the probability of wind speeds of less than 2.5 knots is only 2 to 5 percent. To a layman, this certainly would seem to indicate that the aforementioned wind speed assumptions are quite conservative.
But are they?
The National Weather Service sites in the area are generally located at airports and have anemometers with stall speeds of about 2.5 knots, or about 4 ft./second. Thus, winds speeds of the critical range of less than 2.5 knots are recorded as zeros. To make this data useful for estimation of low wind speeds, all of the data from the sites shown in Table 1 (see next page) for the period of noon to 5 p.m. were first aggregated, and the zero wind speed events were distributed based on Rayleigh distribution. Of the "zero readings," 55 percent were assigned a value of 2 knots (3.4 ft./second), 30 percent a value of 1 knot (1.7 ft./second), and 15 percent a value of 0 knots.
Next, the effect of the wind direction should be considered. A parallel wind has approximately 42 percent of the cooling effect of a perpendicular wind, while winds at intermittent angles have cooling effects between these two values. Because the transmission lines have many angles and go in many directions, wind direction occurrences of 15 degrees, 30 degrees, 60 degrees, and 75 degrees were assumed to have an equal probability of 25 percent. Based on this, an "effective wind speed" distribution was created.
Figure 1 provides several important conclusions for wind conditions during the afternoon of Aug. 14, 2003:
- There was a probability of 5 percent of the effective wind being less than 0.6 ft./second in Ohio. At 0.6 ft./second, the effect of convection is equal only to the so-called natural convection, i.e. zero wind speed.
- Effective wind speed of 2 ft./second had a probability of 18 percent. The 2 ft./second value is commonly used by most utilities as a "safe" wind speed assumption.
- The median wind speed was only 4.5 ft./second, compared with summertime average median value of 12 ft./second, according to Wind Resource Atlas of United States
- The probability that the effective wind speed was less than the 3 ft./ second assumption was 30 percent.
- Some Midwest utilities assume a wind speed of 4.4 ft./second. Note that the probability that the effective wind speed was less than this value was 50 percent! Thus, this can hardly be called a "safe" assumption.
What happens if the wind speeds are less than the assumed values? This can be investigated by using a common ACSR 54/7 "Cardinal" conductor and assuming that the conductor is rated to operate at a temperature of 212 F.1 If the ambient temperature is 90 F and if the wind speed is 3 ft./second, the line would be rated at 1326 A. If the wind speed is instead assumed to be 4.4 ft./second, the line would have a rating of 1451 A. The 1.5 knot (2.5 ft./second) assumption would result in a rating of 1278 A. The figures in Table 1 show the consequences.
Actually, conditions in the line corridors in question were most likely even more dangerous than shown by the above airport-based weather data. Transmission corridors are much more sheltered by trees and terrain than airports. Often, wind speeds in transmission line corridors are found to be only 40 to 60 percent of those at nearby airports.2
How are the line clearances from ground affected? Assume that the lines were designed for 1,000-foot spans with final sags of 33.6 feet. If still air conditions occurred, lines designed with the 3 ft./second wind speed assumption would sag 4 feet more than assumed while lines designed with the 4.4 ft./second wind speed assumption would sag 5.2 ft. more than assumed. For lines using the PJM assumption of 1.5 knots, the sag would be 3.3 ft. more than assumed. Thus, it is very likely that many lines operated under those assumptions under the wind conditions on Aug. 14 would have been significantly below the safe clearances mandated by NESC (National Electric Safety Code), because most utilities apply safety clearance buffers of less than 3 feet nowadays.
What could the consequences have been? In an article in the Cleveland Plain Dealer,3 a utility spokesman described the ground faults as "glitches," stating, "The ability of a line to trip and then immediately reset is a normal function of the power grid."
In a phase-to-ground fault of a relatively short transmission line, such a "glitch" could mean arcs of 10,000-20,000 A, with instantaneous arc power of several megawatts. Within normal circuit breaker operating times, the arc energy can equal several sticks of dynamite.4 The spokesman is fortunate that these "glitches" in Ohio spent much of their energy on splintering trees, instead of, for instance, hitting a schoolbus.
In many cases over the past decade or so, utilities and ISOs have adopted both less conservative rating assumptions and smaller clearance buffers in an attempt to maintain adequate capacity in their transmission systems-all while actual line loadings have increased significantly. In reality, we have degraded the system reliability, and even more importantly, potentially endangered public safety.
Fifteen years ago, it was extremely rare to operate a line at above 75 percent of its rated capacity. Today, such occurrences are commonplace. Thus, when emergency overloads occur, a conductor's temperature can already be very close to its design temperature, or perhaps even above it, giving operators very little time to react to the situation, even if they know about it.5
It is true that, most of the time, actual line ratings are substantially higher than those assumed for emergency rating conditions. But if the operators do not know their line ratings in real time, they must base their ratings on extremely safe rating assumptions.
Without adequate real-time monitoring of the actual thermal state of the lines, overaggressive ratings need to be scaled back. That is bad news for the U.S. transmission network unless real-time ratings are adopted, since otherwise network capacity would be shown to be even worse than what is currently thought. But with optimized use of transmission lines up to their full, measured, real-time capacities, much additional capacity could be gained, along with 100 percent assurance of public safety and system reliability.
- Other assumptions were full sun, and absorptivity and emissivity of 0.8. The actual ambient temperature was assumed to be 87 F. For sag calculations, it was assumed that the conductor had an initial stringing tension of 20 percent and that the conductor stress-strain curve was according to full aluminum compression model, i.e. no birdcaging.
- Tapani Seppa & Dale Douglass: "Safe Weather Assumptions for Line Ratings", Electrical World, January 2001, pp.21-22. (Can be found at Web site www.cat-1.com.)
- "Ohio Power Grid Said to Have 64 Glitches" Cleveland Plain Dealer, Sept. 26, 2003.
- A stick of dynamite has an energy content of 300 kWsec. A 10 kA arc, with an arc voltage of 500 V has an instantaneous energy of 5 MW. For a 100 ms breaker, the arc energy would be 500 kW, i.e. 1-2 sticks of dynamite. Quite a "glitch"!
- Assume that load of the Cardinal conductor used in above example rises from 50 percent of rating to 120 percent of rating (AEP case) when the wind speed is 2 ft./second instead of assumed 3 ft./second. In 15 minutes, the conductor temperature would increase from 132 F to 237 F. If the initial load is 85 percent and the final load is 120 percent, the temperature would rise from 198 F to 262 F. High system preloads increase the danger of contingency loadings.
Articles found on this page are available to subscribers only. For more information about obtaining a username and password, please call our Customer Service Department at 1-800-368-5001.