AltaLink, L.P. is responsible for electric power delivery to more than 50% of the population of the province of Alberta, Canada, providing power to more than 800,000 direct and indirect customers. It owns and operates 11,000 km (6835 miles) of transmission lines at voltages ranging from 69 to 500 kV, with the 240-kV network forming the system's backbone. AltaLink's service area in southern Alberta is characterized mostly by flat open terrain, with frequent steady winds from all directions depending on the season. Many of the 240-kV circuits are on double-circuit, steel-lattice, L-Towers, which were designed in the 1960s and extensively applied in the 1970s and 1980s. There are about 3000 of these L-Towers supporting twin-bundle and single-conductor lines in AltaLink's transmission network. These are relatively light towers and use fewer bracings than the utility's other tower designs (Figure 1).

During a flying inspection in December 1995, a broken hanger of one crossarm of an L-type tower was found (Figure 2). The line was originally constructed using cushioned suspension clamps with helical rods, without additional damping on the conductors. This was later found to be a system that would protect the conductor but could transfer wind energy up the insulator strings to the structure. Consequently, the problem was expected to be limited to lines with this type of construction.

Further inspection showed that the damage was more widespread. In some areas, almost one-third of the towers of this design experienced some damage. The damage varied in severity from minor cracks originating from the boltholes at the tower extremities (Figure 3) to full failures of the upper tension members and release of the suspension assembly and conductor (Figure 2). The failures developed within about 15 years of erection, and occurred on lines independently of their direction, and even on some arms with no conductor installed. Middle arms were revealed to have a slight increase in the occurrence of damage than the upper and lower arms. Linemen often reported hearing noise emanating from the towers and occasionally observed the vibration of some hanger members. Metallurgical examination of the fracture surfaces of failed members confirmed that the failures were the result of fatigue.

At the time problems were first discovered with the L-Tower, the transmission line assets in southern Alberta were owned by TransAlta Utilities Corp. TransAlta sold its transmission assets to AltaLink, L.P. in 2002. The L-Tower Rehabilitation project was initiated by TransAlta, and has continued under AltaLink's direction.

In 1996, TransAlta assembled a team of specialists to find a cost-effective solution to the fatigued tower arms. The team's goal was to reduce dynamic stresses at the critical sections to half of the existing level without increasing stresses elsewhere, which would extend fatigue life by a factor of 10.

A review of previous problems of tower damage in Canadian utilities indicated that the fatigue could be from vibration as a result of either wind-induced aeolian vibration of the conductors or direct action of the wind on the member. The hanger members were 44.5- by 44.5- by 3.2-mm (1.75- by 1.75- by 0.125-inch) galvanized steel angle. These members have slenderness ratios of 296 and 283 for the middle arm and for the top and bottom arms, respectively. This compares to the predicted critical value of 250.

Two L-Towers that had experienced hanger member damage were chosen for field measurements to clarify the cause of the vibration. The two monitoring sites were chosen on lines oriented in the east-west and north-south directions to help identify the winds that caused the vibration.

At the first site, strain gauges and accelerometers were attached to six hanger members, vibration recorders were mounted on two of the conductors, and a weather station giving ambient temperature, wind speed and direction was mounted on top of the tower. An instrument shed at the base of the tower was used to house a computer for storing and transmitting data to the downtown office. At the second site, four vibration recorders were mounted to measure the vibration amplitude and frequency of two conductors and two hanger members. The data from all instruments were sampled for about 10 seconds every 15 minutes. Using binoculars from the ground, the vibration recorder data were collected every few days. The towers were instrumented in November 1996, and vibration recorder data were taken until the end of February 1997 when the second site was. The first site was used in further studies until March 1998.

Both test sites indicated possible high levels of vibration in the hanger members but little significant vibration in the self-damping conductors. Unfortunately, the results could not conclusively identify a principal driving mechanism. At the first site, the strain gauge signals indicated resonance frequencies on the top and bottom arms at 13-16, 25, 33, 39 and 42 Hz, and on the middle arm at 17, 20, 23, 27 and 35 Hz. This series of measurements served to eliminate wind-induced vibration of the conductor as the driving mechanism causing the fatigue damage. The subsequent focus shifted to member vibration caused by vibration of the overhead shield wire. While at surprisingly high amplitudes, shield wire activity was at frequencies higher than that of vortex shedding across the hanger members themselves. It could not be directly linked as a primary driver to the hanger member vibration.

Analytical Solutions

On some towers where failures had occurred, damaged hanger members were replaced with thicker members before an alternative solution was developed. These repairs were relatively costly and needed a power outage for installation. Because of the large scale of the problem and its projected cost of CDN$25 million if applied throughout the TransAlta system, the utility sought a lower cost, yet still effective, solution that could be applied to all towers.

The original designs of the top (Figure 4) and middle tower arms, plus 16 alternative strengthening schemes were studied using modal analysis and finite element techniques. Some of these schemes involved substituting thicker members, adding various bracing designs, and adding gusset plates at the end of the tower arm.

The natural frequencies and stress distributions were determined in the 0 to 50 Hz range. For the original design, the modal analysis predicted 40 natural modes in the frequency range up to 24 Hz. Twenty-eight of these involved motions of the hangers, many of which would result in high dynamic stresses at the end boltholes. Figure 5 shows one of the predicted low-frequency modes, at 13.6 Hz. This study showed that several resonances could be excited by winds in the aeolian range and that, in the original design, there could be high stresses at the boltholes at the ends of the arms.

When the alternative bracing schemes were analyzed, a simple box-bracing scheme for the middle arm met the target reduction of stress. Figure 6 illustrates the altered vibration mode at 13.9 Hz after a similar box-brace was added to the top arm. While it was successful for the middle arm, this scheme only achieved a minor reduction in vibration stress at the bolt location in the top and bottom arms and would not have been successful in prolonging the service life of the tower arms.

Ultimately, finite element studies indicated that use of box-bracing for the middle arm, coupled with a more complex bracing scheme for the top and bottom arms, came close to meeting the required stress reduction. Implementation was estimated to cost $12 million, a marked reduction from that of the initial solution. However, it was still considered sufficiently high to justify further investigation.

The addition of bracing was considered a suitable modification to the design of new towers of this class. Subsequent investigations focused on adding damping to the existing structures.

Experimental Solutions

A full-scale model of the tower shaft and a pair of middle arms was erected in an indoor facility to allow measurement of member stresses and displacements during vibration testing (Figure 7). This followed the pattern used in an earlier study of tower vibration problems in a Canadian utility. This model was intended to identify vibration modes that introduce high stresses at the boltholes, and to evaluate alternative bracing schemes and other means of reduction of stresses in the hanger members. The structure was attached to a heavy steel base plate. Concrete blocks were used to simulate the conductor dead weight, and the excitation was a random signal provided by a servo-hydraulic shaker. Filters removed vibration signals below 5 Hz and above 100 Hz. Strain gauges mounted at the critical sections next to the boltholes on the arm members and accelerometers at their mid points, were used to indicate stress and displacement magnitudes, respectively. The data were stored in a local workstation, and spectral analysis was performed using Labview software. The fatigue damage during each measuring period and the fatigue life of the member were estimated from the measured amplitudes of strain, with analytically derived factors to estimate the peak stresses at the boltholes and a generic fatigue curve for the steel.

In a preliminary test, one Stockbridge damper, normally used on conductors with a diameter of between 23 and 36 mm, was attached to the midpoint of a single hanger member. The vibration level was found to be significantly reduced. This led to the evaluation of a range of damping options, using three sizes of Stockbridge damper from three manufacturers, linking the two hanger members in an arm by various means, including different lengths of steel cable, and attachment at different positions of the hanger. For each alternative, the dynamic stresses at the critical boltholes were recorded for the frequency range of 0 to 100 Hz, and the fatigue life was estimated from these stresses. Some 50 alternative designs were tested, leading to an economic choice using one of the smallest-sized Stockbridge dampers evaluated in the tests.

The final solution used any of the three makes of damper, mounted on a 19-mm (0.75-inch) diameter galvanized wire rope 50% longer than the distance between the attachment holes. The attachment point was at the 30% position along the hanger length. The fatigue life improvement was estimated to be an increase of 15 times the original design. Figure 8 shows the resulting optimized damping solution as installed in the field.

Field Evaluation

To ensure that the damping solution also was effective on a real tower, the first test site was equipped with the dampers mounted in the same manner as in the laboratory trials, on three of the six arms. This arrangement was monitored during the five-month period between November 1997 and March 1998, and the vibration and weather data were compared to equivalent data from the six months between May and October 1997 with no dampers. The weather, strain gauge and accelerometer data were sampled at 15-minute intervals, and the Labview software was set up to display essentially the same data as for the preliminary studies. The cumulative damage corresponding to each signal sample was used to distinguish periods of severe vibration from lower levels of activity. The pick up of several spurious signals, attributed to radar and communications signals from the nearby Calgary International Airport, caused some difficulty. Until all severe vibration signals had been visually examined, it was not possible to separate the true structural vibration from these spurious signals. After this laborious task had been done, patterns of vibration response to wind speed and direction became apparent.

Figure 9 shows the response in terms of cumulative fatigue damage, on a logarithmic scale, of an undamped hanger to wind direction. The figure shows that the hanger vibrated most severely when the winds were from the east to southeast and southwest to west directions. This corresponds to the normal wind acting directly on the hangers, and indicates that conductor and overhead ground wire vibration are not the cause of the fatigue damage.

Figure 10 shows the response of an undamped hanger to east winds, one of the wind directions known to induce hanger vibration, for a six-month period. This figure shows that the vibration tends to be increased by higher wind speeds. Figure 11 shows the corresponding response of the hanger to east winds, for a five-month period with dampers attached. Clearly, the response to the same wind direction with the addition of dampers is much less.

Based on an analysis of the weather records for the test location, especially wind speed and direction, the overall benefit of dampers was estimated to be an improvement of fatigue life by an average factor of 13 on the damped arms. The test tower had dampers on one side only, but the strain gauge data showed that the dampers had a positive influence also on the undamped arms on the other side of the tower. The fatigue life improvement factor for these undamped arms had an average value of 3.5. The hanger member dampers will be applied to all six arms of the towers in need of correction. With dampers on all six arms the projected extension of fatigue life for the fully damped towers is expected to exceed the above estimates for the damped arms.

Damper Endurance

For reliable, long-life service, the applied dampers must last for the required service life of the towers. The spectrum of vibration applied to the dampers in the above tests was used to test the response of the hangers (velocity vs. frequency) with dampers in place. Specifications for endurance tests of Stockbridge dampers were used to establish the fatigue test procedure and acceptance criteria for the arm dampers. The test setup developed by the damper manufacturer for this application supported two dampers together on the shaker. The energy absorption of the dampers was measured individually and as a pair. Tests were performed on two prototype and two production dampers, and both types met requirements for energy absorption and fatigue endurance.

The cause of tower member vibration and fatigue has been identified as direct wind action on slender hanger members. A damper solution to this tower vibration problem has been developed and implemented based on analytical models and laboratory and field testing. The dampers are being installed on AltaLink's 240-kV system at a rate of 20% annually. AltaLink expects the damper solution to save $9 million over the structural reinforcing alternatives commonly applied to this problem.

David Glyn Havard received the BSME degree from Imperial College (London. England) and MS and doctorate degrees in civil engineering from the University of Waterloo (Ontario, Canada). Havard was a senior research engineer at Ontario Hydro Research Laboratories, solving mechanical engineering problems related to the transmission system. He now runs a consulting company dealing with vibration problems of overhead conductors and support systems. dhavard@interlog.com

Owen C. Perry received the BSCE degree from the University of Manitoba in 1977. He was employed with TransAlta Utilities Corp. for more than 20 years working in transmission design, development, standards, and EPC (Engineering, Procurement and Construction). He is currently employed by SNC-Lavalin ATP Inc., in Transmission EPC work. owen.perry@snclavalin.com