hosted by Vicnet
The first King's Bridge, Bendigo, under test.
Photo: University of Melbourne Archives,
Reinforced Concrete & Monier Pipe Co Collection GPNB/1100.
More historic images of King's Bridge may be found on the internet at:
Within UMAIC search under Record ID for: First King's Bridge: UMA/I/6203, 6245, 6246, 6248, 6390, 6391, 6402. Second King's Bridge: UMA/I/6395.
Any enquiries to UMA concerning these images should refer to their Location Numbers: First King's Bridge BWP/23792, BWP/23786, BWP/23789, BWP/23791, BWP/23800, BWP/23801, BWP/23806; Second King's Bridge BWP/23803.
Further images held at UMA have Location Nos. BWP/23784, 23785, 23787, 23788 and GPNB/1099, 1101, 1102, 1252, and 1253.
A separate page on this website covers all eight Monier Arch Bridges built in Bendigo by Monash & Anderson, and includes a brief mention of the construction, partial collapse, and rebuilding of King's Bridge. The reasons for devoting an additional, separate page to the first King's Bridge are:
The photograph above shows an early stage in the testing procedure, as required by the City Engineer. Partial collapse was induced when the traction engine was required to turn round and back up to the steam roller, to concentrate the load of both back axles near mid-span.
Since late 1897 Monash and Anderson's trump card in their drive to establish themselves as designers and builders of civil engineering projects had been control of rights in Victoria to the system of reinforced concrete patented by Joseph Monier. In 1900 they still saw its potential in terms of arches and underground pipes. The Monier bridge was an adaptation of the traditional masonry bridge, with reinforced concrete substituted for brick or stone in the vault of the arch. [Drawings.] The "concrete" was initially a coarse mortar and the reinforcement consisted of grids formed by closely-spaced small diameter rods.
A major opportunity to exploit the patent occurred when it was decided to control flooding and silting in the Bendigo Creek by building a concrete-lined channel through the City, requiring the replacement of a number of bridges. The City Engineer, J. R. Richardson, had prepared designs for these using the conventional technique of steel girders resting on masonry supports and carrying a timber deck. Amongst them was King's Bridge, planned to carry what was then White Hills Road across the 18.3m wide channel at an extreme angle, or 'skew'. Monash and Anderson realised at once that a Monier arch for this particular location would not compete in price with Richardson's version. It was contemporary practice, for design and analysis, to imagine a masonry arch cut into vertical slices and to investigate the theoretical stability of each independently of its neighbours. In the case of a skew arch, the slices were imagined cut parallel to the parapets, along the skew. This made the effective span of King's Bridge relatively large, at about 28.5m. To make matters worse, the need to provide clearance for floodwaters demanded high springings, resulting in a relatively flat arch. This combination of factors would result in large internal forces for which a thick vault and heavy abutments would have to be provided.
M&A therefore asked the City Engineer to treat tenders for this bridge separately from those for seven other more straightforward bridges being invited at the same time. Richardson indicated that he would comply, but without warning called tenders for all eight. M&A appeared to have been eliminated from the running. However, the lowest bid for King's Bridge was £1600 compared with an allocation of £1200, and the partners seized the chance to offer a basic Monier arch for the lower figure. Meanwhile, Monash set in train one of his highly effective lobbying campaigns, and M&A eventually won the contract to build all eight bridges against stiff opposition from the local brick industry and its unions. It is ironic that the partners fought so hard to secure the contract that would bring them so much distress. During negotiations, Richardson had added several stipulations to the standard form of contract. Most importantly he had decided that King's Bridge, situated on the main route north through Bendigo, must be able to withstand the weight of 30-ton boilers which might be transported to mines in the region. He therefore specified that it be tested by the combined weight of a steam roller and a traction engine, both weighing 15 tons. This was double the normal requirement.
M&A exercised the Victorian rights to the Monier patent under licence to a Sydney firm, at that time named Gummow Forrest & Co. It was understood they would assiduously promote and maintain the reputation of the Monier system and that they had a right to, and would accept, GF&Co's advice on technical matters. This came mainly from their chief design engineer, W. J. Baltzer. The archives show that M&A performed parallel calculations for some bridges (and sometimes reached slightly different conclusions) but there is no record that they did so for King's Bridge at this early stage.
Early tender drawings showed the arches to scale, but their thicknesses were not specified in figures and were probably based on typical sizes found in the Monier handbooks and technical literature.
At the date of the contract, the profile of the arch and its thicknesses at springings and crown appear to have been determined solely by GF&Co. Before construction commenced, M&A persuaded Richardson to reduce the angle of skew from 53° to 50°. This not only simplified construction, but reduced the span and hence the forces involved. M&A made necessary adjustments to the profile, but proposed to leave all other dimensions unaltered.
The angle of skew quoted here is the angle by which the centreline of the bridge deviates from that of a 'square' bridge - one that crosses a creek in the most direct manner possible. Contemporary local practice was to refer to the complementary angle, resulting in a paradoxical statement that the skew had been "reduced from 37 to 40 degrees".
Early designs for King's Bridge assumed that rock would be found at a depth of three feet, but by the time contracts were signed, it was required that foundations be carried down seven feet (2.13m) below the creek bed "except rock is reached". Excavation and concreting were carried out under awkward conditions with pumps racing to overcome an inrush of water and fine silt penetrating the timber cofferdams. The hurried nature of this dangerous task resulted in clashes between M&A and Richardson who tried to insist on his right to see the exposed foundation material, and approve it as adequately strong, before it was covered with concrete. Rock was found on the downstream end of the right abutment at about seven feet, but it was decided to base the upstream end of the right abutment and the entire left abutment on "stratified pipe clay" at about this depth. By 17 January 1901 both abutments had been concreted to springing level.
About this time M&A sent their foreman instructions, based on fresh calculations, to reduce the thickness of the arch slightly below the values specified by GF&Co, especially towards the edges which carried the spandrels. It is doubtful whether these instructions were carried out. The only slender evidence found in confirmation is a passing remark in a site diary: "King's: altering side boards". (These were the strips of vertical formwork along the edges which determined the thickness of the concrete.) However, there is no evidence that the change was discussed with either GF&Co or Richardson, or ever approved by them. Since even minor changes were often fought over at great length, it is difficult to believe that M&A would have attempted to act unilaterally in such an important matter; or that Richardson would have allowed the move to go un-noticed and unchallenged. All evidence given at the inquest was that the dimensions had exceeded those shown on the drawings. The reference to a reduction in thickness must therefore remain a mystery.
The forming of the arch itself was to be done in three parallel strips, each cast in a single day from one abutment to the other. Timber frames, or 'centres', were erected to support timber sheeting which defined the shape of the underside ('soffit') of the arch. On this was placed the lower reinforcing grid, supported a couple of centimetres above the sheeting by small briquettes of hardened mortar. The coarse mortar for the arch, with a consistency like "damp earth", was then pounded into place around and on top of the steel rods. When an appropriate thickness was reached, the upper grid was laid on top and covered with a final layer of mortar. A light-hearted newspaper account of this process is reproduced in Section 6 of the Dossier (entry 00/11/24). The first strip, on the downstream side of the bridge, was cast on 8 February 1901. Some difficulty was experienced in removing the centres from beneath this strip, suggesting that it had sagged more than anticipated. The process was repeated to complete the other strips by 1 April. Construction of the downstream spandrel wall had started while the first strip was still supported by its centering and was probably complete before the upstream strip was cast. As the second spandrel rose, earth fill was compacted between the walls to build up the road. It is probable that the spandrels were built in concrete, though the parapet walls were brick. Later in April Monash complained to his foreman that the bricklayers had been "fooling about" for several weeks and expressed his impatience to get traffic consolidating the fill because he had asked Richardson to conduct the load test in mid May. The partnership was having cash-flow problems and full payment would not be received until the bridge had been judged satisfactory.
On 14 May Richardson and a team of Council employees assembled to conduct the test. Anderson was engaged on other work and Monash represented the partnership. The council workers installed simple instruments (extensometers) under the bridge to measure deflection due to the test load. Richardson first instructed the Council roller driver to compact the newly laid road surface. The roller made twelve passes, but avoided the extreme northern (downstream) edge where a water main had been laid within the fill. The roller and traction engine were then both placed on the bridge so that Monash could take photographs. After this, preparations were made to conduct the formal test, with council workers going under the bridge to record the readings of the extensometers. It was then noticed that a fine crack had developed in the soffit of the downstream strip near the crown, running across its full width and extending into the centre strip. There were signs of distress in other locations and the extensometers were already indicating some permanent deflection.
Richardson, Monash and others examined the bridge; but Richardson decided it was safe to proceed with the test. There is no record of Monash having made any attempt to dissuade him. This may seem strange, but both engineers would have known of European tests of prototype Monier arches in which the load required to cause failure had been three times that at which the first signs of distress appeared. Monash was convinced that all cracks were superficial and that the bridge would never again be subjected to anything like its present load. As the test continued, the roller and traction engine were brought close together near mid-span on two occasions while readings were taken from the extensometers by men standing under the arch. After some consideration, Richardson decided that the test had not been sufficiently severe, as most of the weight of the engines was concentrated on their back axles and so far they had been placed in tandem. He ordered the traction engine to leave the bridge, turn round, and back up to the roller near mid-span to bring its rear axle close to that of the roller, thus simulating his 30 ton concentrated load.
Monash had argued, and continued to do so, that this made the test unduly severe, as the weight of a 30-ton boiler would be spread over many axles.
The machines were placed mainly over the middle strip, but each had two wheels on one side resting on the downstream strip. The diagram below, adapted from one prepared by Professor Kernot, shows the three strips spanning between the west and east abutments, the positions of the traction engine and roller, and the portion of the northern strip which collapsed (shaded).
Plan view of collapse, based on a sketch from the University of Melbourne Archives, Reinforced Concrete & Monier Pipe Construction Co. Collection.
As the council workers took up positions under the bridge for further readings it was noticed that deflection was increasing at an alarming rate. Pieces of mortar could be heard dropping from the crack and splashing into the water. There were shouts of warning and a scramble to get clear. Those on top of the bridge felt the downstream strip starting to give way. The crew of the traction engine managed to jump clear of their machine, but descended with the arch as did a group including Richardson, which had been on top of the bridge. The concrete of the downstream strip separated from that of the middle strip, leaving a clean face. The lower lateral reinforcing rods, which had extended some way into the middle strip were ripped out, while the downstream spandrel wall and parapet were thrown into the channel. The roller remained precariously perched on the edge of the middle strip; but the traction engine fell sideways into the channel. Its fall killed A. E. Boldt, a business associate of its owner, who had been next to it looking over the parapet. Surprisingly, considering the jumble of concrete, bricks, coping blocks and men that had been projected through the air, he was the only casualty.
University of Melbourne Archives BWP/23802, Reinforced Concrete & Monier Pipe Co Collection.
Reactions to the disaster and attempts to explain it are covered in some detail in the newspaper reports (Dossier, pp.13-18). Monash and Anderson were mystified by the collapse. They were sure of their calculations and of the quality of materials and workmanship. Anderson wondered if there had been a layer of soft foundation material, concealed below the depth excavated, which had allowed the abutments to move apart. Tradesmen, accustomed to masonry bridges, suggested the arch may have been too flat. Others, especially those whose business was threatened by the Monier system, suspected the quality of construction. In commercial terms the situation was critical, as the disaster threatened to bring the Monier system into disrepute throughout Victoria and deprive M&A of future highly profitable work. The partners could only assume that some hitherto unsuspected technical problem was to blame, and retained W. C. Kernot, Professor of Engineering at the University of Melbourne, to investigate on their behalf. One possibility was that there was something wrong with the above-mentioned conventional method of analysing skew arches, even though this carried the imprimatur of the doyen of 19th century British engineering theory, W. J. M. Rankine.
In searching for a more rigorous alternative, Kernot consulted leading mathematicians, but it was agreed that the mechanics of the skew arch were beyond their capabilities; certainly in the time frame set by the inquest. Attention then turned to small-scale physical models, to gain some understanding by observing their deformation under load. Kernot made models in cement, plaster and even soap, but the most successful were rubber models, one of them a sheet of rubber suspended so as to adopt the inverted form of the arch. One of Kernot's brothers, W N Kernot, a lecturer in electrical engineering at the Working Men's College, studied the flow of electricity through a metal sheet cut to the shape of the arch, exploiting the fact that the equations governing distribution of electrical potential are similar to those governing intensity of stress.
The Working Men's College became the Royal Melbourne Institute of Technology and is now RMIT University.
Although they could not give accurate figures, these studies showed that in a skew arch the internal thrust would tend to concentrate along the shorter diagonal (looking in plan). In the case of King's Bridge, with its high angle of skew and high span-to-width ratio, the actual stresses would be more than four times those calculated by the conventional method. In a private communication to M&A, Kernot mentioned that another of his brothers, M E Kernot, a senior engineer with the Victorian Railways, had earlier studied the behaviour of a proposed highly-skewed masonry arch and had reached similar conclusions. Unfortunately, when the project was abandoned, he had shelved the theoretical problem and had been too busy to publish his partial findings. Observation of the ruins of King's Bridge and tests made on samples of material suggested that while the concrete of the arch could withstand even the heightened stresses, the abutment concrete, with its lower proportion of cement, had been unable to do so and had been crushed, allowing the arch ring to descend. Another factor was that the abutment, seen in plan view, possessed a sharp projecting corner which would be unequal to the task of supporting the concentrated thrust, tending to act along the shortest diagonal.
Kernot presented this evidence to the inquest with the aid of a blackboard and assured the jury that every text-book he could lay his hands on, including the great Rankine's, advocated the conventional approach. All witnesses testified to the high quality of workmanship and material, implying that the fault could only lie in the inadequacy of the theory employed. A design engineer who, in good faith, uses currently accepted theory in a situation where it proves inadequate has a strong but not unassailable defence in law against imputations of negligence. The Bendigo jury evidently decided that the designers and builders in this case had acted reasonably and could not have been expected to foresee the problem. The verdict was that Boldt had met his death accidentally, "and that no blame can be attached to anyone".
This was not to be the last word on the subject. Others continued to look for further factors which might have contributed to the collapse. The Chief Engineer of the Public Works Department of Victoria, Carlo Catani, noted the existence of cracks on the still-standing upstream side of the bridge, affecting the arch, spandrel and parapet. Catani had been supportive of M&A in their efforts to introduce the Monier system, but there was some friction when he drew these signs of distress to their attention. Anderson and Kernot asserted the crack was not as extensive as Catani claimed and that when Kernot had attacked it with a crowbar it had proved to be superficial. A correspondent to the "Bendigo Advertiser", signing himself "Ratepayer", remarked that Kernot's similar attack on the abutment concrete with a pick-axe showed him a weakling, as much of it could be dug out with a knife. "Ratepayer" claimed that the centre portion of the previously-completed Oak Street arch was level over a distance of 4.5 feet (1.4m) and that King's Bridge had suffered from a similar defect. In reply, a correspondent signing himself "No Axe to Grind" derided these statements and claimed that "Ratepayer" was a local personality with a vested interest in timber bridges.
Prejudiced assaults and disinterested technical hypotheses were equally quashed by M&A who now turned to the challenge of re-designing the bridge. The cause of failure had been identified as the inability of the abutment concrete to resist the magnified stresses imposed by the skew arch, so there could be no question of re-building the arch in its original form. This would have required demolition of the abutments and their reconstruction in stronger concrete, and/or a large increase in the thickness of the arch near the overstressed corners to reduce the intensity of stress. Both these measures would have been expensive, and would have involved the designers in a situation still not fully understood. M&A therefore decided to retain the existing abutments and build a new pier in midstream so that the bridge would consist of two arches each of 13.2m span rather than one of 28.5m. This would reduce the calculated thrust on the abutments to little more than a quarter of the value for the single arch. Also, the ratio of skew-span to width would be reduced by more than half, and the abutments of each arch would be situated more directly opposite each other.
King's Bridge reconstructed with a central pier (barely visible) and two spans. University of Melbourne Archives BWP/23803, Reinforced Concrete & Monier Pipe Construction Co. Collection.
Bridge coordinates: -36.74364, 144.29165
The process of building the pier and the new superstructure was reasonably straightforward and is covered on the Bendigo Bridges page in this website. What is noteworthy is the treatment of Kernot's findings. He presented them to the Victorian Institute of Engineers on 7 August 1901, where they were well received; and published them in the New York "Engineering News" on 11 June 1903 with the intention of reaching an international audience. However, Baltzer was not convinced. He argued that as the downstream strip had separated from the rest of the bridge, it must have acted much like the narrow slice of traditional theory, with a span of 28.7m. He thought failure had been due to a horizontal sliding force derived from the oblique angle at which the thrust met the abutments.
This had caused the abutment concrete to fail as a result of shearing, rather than crushing stress. His calculations for the replacement bridge employ traditional theory based on the skew span, with no allowance whatsoever for Kernot's magnification of stress. Monash adopted the same approach in checking stresses due to the steam roller, shortly before the new bridge was tested in January 1902. So, although Kernot had convinced the inquest jury, the engineers concerned with re-design rejected his findings! Who was correct?
There can no definite answer to this question. The power of modern computers would allow us to represent the underlying rock and soil, the abutment blocks, the complex shape of the skewed arch with its varying thickness, the spandrel and parapet walls, and finally the earth fill in the form of small three-dimensional blocks; and solve thousands of equations expressing the balance of forces between them, or the compatibility of their deformations. However the original arch concrete, spandrels, parapets and fill have all disappeared, so we shall never know exactly their mechanical properties and whatever bond or friction might have developed between them. Stresses in flat arches are highly sensitive to geometrical imperfections in the curve of the profile and we cannot know what these were. We could still determine the actual mechanical properties of the ground beneath the existing abutments, but this would be extremely expensive. Thus, even the most rigorous modern analysis practicable could only give us an indication of the likely cause or causes of failure.
My hypothesis is that in the previous experience of GF&Co and M&A with Monier arches, certain computational procedures, established by long experience with masonry bridges and sanctioned by experts had been found satisfactory for design and analysis; while certain levels of geometrical accuracy had been found acceptable in construction. When these were applied to the unusual and unsuspectedly complex case of King's Bridge they proved inadequate.
It is possible to obtain some hints from modern research into the traditional masonry arch which may be described as first cousin to the Monier arch. Thousands of masonry arches are still in use throughout Europe and in recent years the growing weight of goods vehicles has stimulated extensive (and expensive) research into their mechanics, including high-powered computer analyses and full-scale tests to destruction using modern instrumentation.
Even so, the mechanics of masonry arch bridges are still not fully understood. See e.g. Page, J. (ed). "Masonry arch bridges", HMSO, London, 1993; and Harvey, B. "Some problems with arch bridge assessment and potential solutions", The Structural Engineer, 7 February 2006, 45-50. In a brief review of the literature in 1999, I could find no report of an experiment that simulated all the characteristics of the King's Bridge test: high skew, point loads at mid-span, spandrel walls and earth fill. However, some studies included one or more of these characteristics and their findings are discussed in Section 10 of our Dossier.
There is no doubt that the concentration of force along the short diagonal occurs as Kernot predicted. It is accompanied by significant twisting of the arch rib: a feature he did not describe. Typical failure patterns of masonry bridges suggest that Catani was not mistaken about distress in the spandrel on the upstream side. They explain the statement of witnesses that part of the parapet "reared up" as collapse took place. These studies also show that a skew arch is stiffer than a square arch, deflecting less under a given load. An arch as flat as King's Bridge may fail by 'snap-through buckling' in which shortening of the arch under compression; or a localised crushing of material; or a slight spreading of the abutments allows the arch to flatten and fall. One or more of these factors could explain why collapse occurred so suddenly and at such a low load, compared with experience of prototype square arches having a higher rise-to-span ratio.
Since the Dossier was written, computer analyses of the arch have been carried out by final-year students in the Department of Civil Engineering at Monash University under the guidance of Dr Riadh Al-Mahaidi, a specialist in finite-element analysis. The assumed properties of concrete and reinforcement were based on typical values found in the M&A archives. These confirm the increased levels of stress demonstrated by the Kernot brothers, but show that if the cross-section and profile of the arch were as shown on the drawings, and the concrete of the arch was as strong as typical cube tests indicate, failure could not have occurred in the arch itself. The computer model could be induced to fail only by permitting large outward movement of the abutments, of the order of 100mm.
Note 1: The above account was extracted from our Dossier on the First King's Bridge, Bendigo, and edited for presentation here.
Note 2: An error in quoting William Julius Baltzer's first name was carried over into several dossiers. It appeared on page 26 of this one.