TV mast safety - high level research!
TWI Bulletin, July/August 1989
Martin Ogle, BSc(Eng), ACGI, PhD, CEng, MICE, MWeldI, is Principal Design Consultant in the Engineering and Materials Group.
Fifty million viewers depend on the Independent Broadcasting Authority's (IBA) network of high masts for their TV pictures. Commercial television is big business and the loss of transmission from any one of these masts can affect the buying habits of more than two million people. These masts are also used to transmit signals from a host of other users including the BBC and the Post Office. It is not surprising, therefore, that the IBA give the maintenance of their structural integrity a high priority.
These masts are now about 30 years old and have reached their original life expectancy. Many of the older aerials are in the process of being replaced and a decision had to be made about the future life of the masts. These had to be guaranteed for the life of the new aerials. It was for this reason that the IBA approached The Welding Institute for technical assistance.
Their prime concern was for the long-term endurance of the main joints. The masts are typically 230 and 300m high thin latticed columns supported by steel guy ropes ( Fig.1). The country is served by a network of about 500 masts and towers, many of which are sited in very exposed locations on the tops of hills or mountains. They are continually buffeted by the wind, sometimes in conjunction with snow and ice.
Fig.1. Sandy Heath mast reaches to a height of 230 metres
The potential effects of these elements on guyed masts have already been clearly illustrated. In 1969 the 380m high Emley Moor mast in Yorkshire collapsed under conditions of wind and ice ( Fig.2). In 1985 the 300m Teutoburger Wald mast collapsed in West Germany under turbulent but not severe wind conditions ( Fig.3). Whilst both these guyed masts were mainly of tubular rather than lattice construction, both failures involved failure of a welded joint.
Fig.2. The lattice section of Emley Moot mast lies in the mud after its quarter-mile plunge to earth
Fig.3. The 15 year old Teutoberger mast lies broken in a German forest (photo courtesy of New Civil Engineer)
Mast design
The masts consist of three circular solid steel legs with welded gussets and bolted angle bracing ( Fig.4). The leg sections were fabricated in about 6m lengths with welded end flanges ( Fig.5). All site connections were bolted.
Fig.4. Typical lattice mast construction with solid round legs connected by welded and bolted flanges
Fig.5. Details of leg to flange welds
It was the integrity of the welded flanges which caused the greatest concern. The original design had been based on a static wind load from BS CP3 Chapter V (wind loads on buildings). This loading was never severe enough to overcome the dead load compression in the joints, thus the welds and bolts were nominally sized, the main load being transferred between the sections in end bearing. Subsequent research showed that this code underestimated the true forces that would exist in this type of structure, particularly when dynamic effects were taken into account. Once the joints go into tension, both the welds and the bolts become vital to the safety of the structure.
Subsequent design checks
Subsequent check calculations were carried out by consultants Flint and Neill Partnership on behalf of the IBA, using later revisions to the wind code, which recommended higher design wind speeds. This showed that a significant number of joints would be expected to go into tension. Some were calculated to be loaded above their stated tensile capacity. In this case special friction clamps were installed to bypass the flange joints altogether. However, this left a significant number of joints on some of the masts which were estimated to be at risk from eventual failure by fatigue. Flint and Neill were fully aware of the difficulty of estimating the fatigue life under such conditions by calculation alone. Not only was the structure as a whole very complex to analyse under aerodynamic conditions, but also the fatigue strengths of the complex leg joints were another uncertainty.
At this stage The Welding Institute was brought in to assist with obtaining more reliable data from real masts and real joints.
Experimental programme
The experimental programme was started in 1979, and has covered the following work:
- Fatigue testing of full scale leg joints and bolts and examination of weld quality.
- Recording of dynamic stress histories in leg joints on three masts during storm conditions.
- Writing of software for fatigue analysis of stress records.
- Development of a technique of measurement of residual forces in mast legs and measurement on a mast.
- Development of a bolt tension meter and measurement of tensions on a real mast.
- Investigation of NDT techniques for leg to flange welds.
Fatigue tests
Very fortunately, a section of unused mast, dating from the original period of construction was available. This enabled an endurance (S-N) curve to be obtained from three test joints in the knowledge that there had been no prior fatigue damage. The tests were carried out in The Welding Institute's 200t Losenhausen machine (see Fig.6 and 7). Some care had to be taken to ensure that the bolts were torqued uniformly so that a purely axial load was applied.
Fig.6. A full size leg joint being fatigue tested in The Welding Institute's 200t Losenhausen machine
Fig.7. Fractured welds in a slip flange joint after test
This S-N curve was used by Flint and Neill for their subsequent calculations.
Stress remedies
Three masts have been monitored so far. The first was the Sandy Heath mast in Bedfordshire. This was chosen for its convenience to Abington so that a trial system could be set up and tested.
The 230m mast at Dover was then chosen for a longer period of monitoring, using the same basic system as at Sandy, but with automatic data collection.
In both cases the legs had to be strain gauged at a height of about 180m above ground. This was no mean task on this open type of structure with no weather protection whatsoever. Fortunately, the necessary agility and coolheadedness of a mountaineer, stamina of an arctic explorer and dexterity of a watchmaker happened to be combined in Brian Martin of The Welding Institute's stress laboratory (see Fig.8).
Fig.8. Hanging on at 190 metres, whilst fixing strain gauges
Access was achieved either by a cage hanging on a single temporary rope or via a single flight vertical 600 rung ladder! Assistance was provided by one of the IBA's own maintenance riggers, who installed anemometers and wind direction indicators nearby.
The strain gauge and wind instrument signals were transferred to the recording instrumentation on the ground by multiway screened cable. This consisted of a UV recorder and seven channel tape recorder in parallel. The three strain and three wind data channels were recorded automatically every time the high level wind speed exceeded a threshold value such as 70 or 80 mile/hr (110-130 km/hr). It was decided to retain the full stress history rather than do real time analysis.
The latter would have resulted in greater instrumentation cost and also involved loss of the basic history. The instrumentation system was designed and maintained by Bob Piggott, now of the Engineering Department's new Structural Section.
The tape recorders were digitised back at Abington and analysed using a special suite of programs written by the Computing Department. This enabled mean stresses, mean wind speeds and directions, maximum three second gusts, r.m.s. stresses, cycle counts and frequency spectra to be computed for each storm. This information was passed on to Flint and Neill who built up a total life projection of fatigue damage using meteorological records.
In 1985 the system was transferred to Moel-y-Parc mast in North Wales which has differences in geometry and is also in very different terrain. A typical example of the data record during a storm is shown in Fig.9.
Fig.9. Typical record from Moel-y-Parc mast during a strong gale
Residual stresses
One of the uncertainties in the fatigue assessment was the possible presence of bending stresses in the leg joints. This would have the effect of premature opening of the joint on one side. The other unknown effect was that of lack of straightness in the length of the mast. This could cause a locked-in bending moment in the whole mast, again resulting in premature axial tension in one or two legs.
An opportunity was taken to obtain information on these effects when the original lattice mast at Wenvoe in South Wales was being dismantled in 1983. Using a purpose made gauge bar, a series of pop marks were made round each leg at three different levels up the mast. The 100mm gauge lengths were measured using a very accurate extensometer ( Fig.10) prior to dismantling. The pop marks were covered up for protection.
Fig.10. Brian Martin makes extensometer readings to estimate the dead load stress pattern in the legs
After dismantling of the mast this section of leg was sawn off to convenient lengths and sent to Abington where the gauge lengths were remeasured. The results showed that the bending stresses in each leg due to slight angular distortion caused by the gusset welds gave rise to bending stresses which were of the same order of magnitude as the axial stresses ( Fig.11). This has an important implication for fatigue.
Fig.11. Effect of gusset weld shrinkage on dead load stress distribution in leg
Bolt tensions
A special bolt meter was developed at Abington to measure the tensions in the flange joint bolts, as the precompression of the joint has a theoretical influence on the weld stresses.
The bolt meter is mounted on pop marks on the ends of an installed bolt and a reading taken. The nut is then slackened and another reading taken. The nut is then retightened back to the original value with the bolt meter still inposition. This technique was used on Mendlesham mast in Suffolk.
Each bolt was done in turn, so that the integrity of the joint is not affected. The bolt meter was calibrated using a pressurised Pilgrim nut, Fig.12, and a bolt of the same dimensions.
Fig.12. Mounting The Welding Institute's bolt meter on to a flange bolt prior to measuring its tension
NDT
A sample joint from one of the Wenvoe legs was taken and a 0.15mm wide crack introduced into the weld by electrodischarge machining (EDM) using The Welding Institute's equipment. The crack was then closed up by peening the surface. Tests were done on this sample to evaluate various NDT techniques for finding root cracks.
Conclusions
The results of this work have enabled the IBA to rank their structures in order of susceptibility to problems of fatigue; to extend their service life; and plan economic strengthening measures where necessary with more confidence than was previously possible without real data from actual structures.
This is another example of successful collaboration between Research Member, design consultant and The Welding Institute where the potential cost savings substantially outweigh those of the investigation.
