Martin Ogle is Principal Design Consultant in the Engineering and Materials Group. He joined the Welding Institute in 1978 after a career in consulting engineering, where he was involved in the design, contract supervision, checking and repair of civil engineering structures. These included a wide range of steel and composite bridges, most notably the cable-stayed Batman Bridge in Tasmania where he was Deputy Resident Engineer for four years. He was responsible for much of the drafting on BS 5400 Part 10 Fatigue of bridges, including the creation of a new fatigue loading specification.
At the Institute he was Head of the Structures Section until 1985. He has been involved in welding design problems on bridges, safety barriers, buildings, masts, power plant, offshore platforms, cranes and machines. One of his roles is to ensure that the Institute's many assets - including laboratory testing, theoretical analysis, and on-site inspection and monitoring - are used to full advantage in solving Industrial Members' problems.
He has been active in drafting design standards for welded steel and aluminium structures for national, European and international application. He has also drafted practical workmanship and quality standards for contract use in the building, bridge and transportation industries.
The design rules for steel bridges needed revising to allow greater economy in construction. Supporting data were provided by testing a scale model bridge to destruction. Martin Ogle explains how TWI made the model and completed the test.
Bridge design trends
Steel plate girders are becoming more and more popular with highway bridge designers. Having spent some years increasing efficiency, the steel supply and fabrication industries are now reaping the rewards.
Steel has always been the obvious choice for long span bridges on account of its strength-to-weight ratio. Since the war however, when steel was scarce, the most popular method of construction for medium and short span bridges has been prestressed concrete. One of its perceived advantages was that it needed no maintenance, whereas steel has always had to be regularly maintained.
This situation is now changing. There is increasing concern about the durability of many concrete bridges, some of which have been found to have seriously deteriorated after only 25 years - less than a .quarter of their 120-year design life. The problems have ranged from poor workmanship (inadequate compaction or insufficient cover to reinforcement) and attack by road salts to adverse chemical reactions between modern cements and certain types of aggregate. [1] The views of many designers and agent authorities on maintenance have therefore changed. It is often seen to be better to accept a known regular planned maintenance expenditure than to run the risk of accruing obviously expensive, but unquantifiable, costs in the future.
Design codes
The efficiency of steel bridges is very much dependent on the standards used to design them. In 1980, a major new limit state code BS 5400 was produced, which took account of the very extensive programme of research work, particularly into buckling design, which followed the collapses of box girder bridges around 1970. Part 3 of BS 5400, which gives extensive rules for assessing the static strength of steel members, represented a large step forward from its forerunner, BS 153.
A great deal of the research was concerned with buckling of individual elements such as webs, flanges and stiffeners and built-up stiffened panels such as diaphragms. Since then, there has been an increasing interest in overall lateral buckling of I-girders, a feature which was a non-problem with torsionally stiff box girders. The Department of Transport has become aware that the rules dealing with this mode of failure in BS 5400 Part 3 are in need of revision, and that greater economy could be achieved.
New research contract
In 1988 the Transport and Road Research Laboratory (TRRL) awarded a contract to consultants Travers Morgan to carry out a study into this subject, commonly referred to as the 'U-frame' problem. This contract was awarded by competitive tender; joint collaborators were The Welding Institute and The University of Leeds. TWI's role was to manufacture and test to destruction a tenth-scale model of a bridge - a slight departure from TWI's usual role which has been more concerned with the fracture and fatigue rather than the buckling limit states.
The results of this test enabled the University of Leeds to calibrate their finite element model of the same structure. This gives the necessary confidence for their parametric study of factors affecting lateral buckling.
The work is now nearing completion. The rest of this article describes the way in which the test was carried out at Abington.
The principle of U-frame action
U-frame action arises when plate girder compression flanges are not supported laterally by transverse bracing. In this case the vertical web stiffeners may have to be used instead, rigidly connected to a cross member to provide sufficient stiffness. If they are not, stresses in the compression flanges of the main girder have to be severely reduced - which results in unnecessarily heavy girders. The basic mode of U-frame action is shown in Figure 1. In Europe in the last century there were many failures of this type of railway bridge because designers were unaware of the importance of U-frame stiffness. The basic problem was eventually explained a hundred years ago by Professor Jasinsky, the celebrated railway engineer. [2]
a) Construction of through-deck bridge
b) Cross section at midspan showing U-frame
c) Buckled shape
d) Transverse bending moment pattern in stiffeners and cross girder
In highway bridges, the most usual form of construction is to support a reinforced concrete deck above the main girders - see Figure 2. In this case the unsupported flange is in tension at midspan so the instability problem does not arise with simply-supported spans. However, many highway bridges are multispan so that the critical region is over the supports.
a) Cross section
b) Elevation
c) Longitudinal bending moment pattern in main girder
An obvious solution to the problem which has been used in the past is to brace the compression flanges together. However, in modern bridges this is considered undesirable - firstly, it adds to cost and maintenance; secondly, the bracings are usually very slender and prone to vibrate; thirdly, they seriously inhibit access for inspection and maintenance equipment.
Design of the test model
The test model was designed, not so much to resemble a real bridge in every detail, as to arrive at a geometry which was particularly sensitive to U-frame action. The geometry finally selected by the three partners in the project and agreed with TRRL is shown in Figure 3.
The steel of the main girders was Grade 50 sheet 4mm thick. The cross members were 15mm square bar in 43A. By cantilevering the girders beyond the main supports it was possible to put the bottom flange into a state of constant compression along its length. This provided the most severe conditions for U-frame action. One of the difficulties in designing a test model of this type is to ensure that the many other unwanted buckling and fracture modes of failure are inhibited from the test without making the model unrepresentative of real conditions in a bridge. Particular care had to be taken to ensure that the restraints necessary to inhibit these modes did not interfere with the required articulation of the model.
The problem with a tenth-scale model is to simulate the welds sensibly. On a real bridge the web-to-flange welds might typically be continuous 6-10mm fillets. It would be impracticable to lay down 0.6-lmm fillets on the model; intermittent 2mm fillets were therefore proposed, which gave a correct scaling of the weld deposit.
Building the model
The main problem to be overcome with the manufacture of the model was dimensional control, which had to be within the tolerances laid down by Part 6 of BS 5400. This required precise cutting of parts and jigging together prior to welding. TIG welding using a 1.6mm A17 wire was used throughout. Flanges and webs had to be butt welded in to full lengths at two positions prior to assembly into girders. Assembly, jigging and welding were carried out by Brian Bartlett of the Arc Welding Department.
Each girder was made up in one length, using light tack welds. The cross members were then tack welded between the two girders to give dimensional stability, after which the main welding was carried out in a carefully controlled sequence. The results were certainly up to original expectations and it seemed a pity that such an excellent work of art was destined to have such a short life!
The test rig
The Welding Institute's 1.2m square box girder strongback, originally used for work on largescale pipe bend tests, was used for restraining the forces on the model, see Figure 4.
The supports and loading arrangement were designed and made by Bryan Martin of the Engineering Department Stress Laboratory. The loads were applied to heavy cross beams at each end of the cantilever arms by hydraulic jacks in series with 100kN load cells, see Figure 5.
The model was supported on rollers to give complete freedom to strain longitudinally. The bearing plates were designed with Teflon pads and a central pivot to allow freedom of rotation of the bottom flange in plan, see Figure 6.
The test
There was some trepidation on the day of the test. A number of visitors had come to witness it. With only one very expensive test piece, there had been no possibility of a dress rehearsal. Would the equipment all work first time? Would the failure mode be the required one?
The load was taken up in increments, and deflections recorded at each stage. The beam curved noticeably in elevation and buckles started to develop in the bottom flange, see Figure 7. As the buckling pattern changed and the load increased, the reserve of strength under severely distorted conditions was impressive to see. There was no sudden snap-through unloading, characteristic of many buckling modes. This gave one a new confidence in the robustness of this form of construction. Plenty of warning of imminent failure is a useful feature of any structure - hence the extreme caution given to protection against brittle fracture which occurs without warning in a mere millisecond.
The final buckling mode is shown in Figure 8. Two intermediate nodes formed in the bottom flange of the main span, the web stiffeners tilting in and out accordingly. The cross girders bent in unison, the moment connection welds holding on to the last. Brian Bartlett and Bryan Martin heaved a sigh of relief as their work passed the final test.
Conclusion
The results are now being studied by Leeds University and compared with their theoretical models. Eventually, this work may be used to formulate new rules for BS 5400 for U-frame design.
This project is an illustration of TWI's resources and adaptability when it comes to specialist tasks which are not part of everyday weld testing programmes.
In these days of computers it is easy to rely entirely on sophisticated theory. The real engineer often needs more convincing. The only way finally to prove the strength of something is to break it! It is hoped that this will encourage other Research Members to make as much use as they can of the facilities at Abington, however strange the project may be.
References
| N° | Author | Title |
|
| 1 | Department of Transport: | 'The performance of concrete in bridges - A survey of 200 highway bridges'. HMSO, April 1989. | Return to text |
| 2 | Timoshenko S P: | 'History of strength of materials'. McGraw-Hill, 1953 | Return to text |
