26 TheStructuralEngineer › May 2015 Project focus Load testing a historic structure Grand Parade, Bath: in situ load testing of a historic structure to avoid unnecessary intervention Kim Collins MEng, CEng, MIStructE, Conservation Accredited Engineer and Associate, Integral Engineering Design, Bath, UK Margaret Cooke BSc, CEng, MIStructE, Conservation Accredited Engineer and Director, Integral Engineering Design, Bath, UK Introduction Bath is a city in southwest England, famous for its creamy limestone, 18th-century architecture and hot springs. It is the only destination in the UK to have the whole city designated a World Heritage site. The River Avon runs through the centre of the city. Over time the general street level has become gradually higher as each layer of civilisation built on top of the last. On the west side of the river – the major historic centre of the city – street level now lies some 8m above river level. This case study concerns Grand Parade, the street running along the river just south of the world famous Pulteney Bridge. Grand Parade is subject to heavy pedestrian and vehicular traffic, particularly during the tourist season. Pedestrians and vehicles are protected from falling into the river by a masonry balustrade approx. 1.3m high, which is listed Grade II (Figure 1). Tourists lean on the balustrade in order to see Pulteney Bridge and its weir below (Figure 2), as do W Figure 1 Grand Parade balustrade E Figure 2 Grand Parade balustrade with Pulteney Bridge and weir Synopsis Evidence suggests that modern loading codes often significantly overestimate real loads and that modern calculations may not be an appropriate way to assess the capacity of historic structures. This article describes how in situ load testing of the Grand Parade balustrade, Bath, was performed to confirm its actual capacity and avoid the need for an intrusive strengthening scheme proposed by the local highways department which would have caused significant damage to the historic fabric.
Transcript
26 TheStructuralEngineer
›
May 2015
Project focus
Load testing a historic structure
Grand Parade, Bath: in situ load testing of a historic structure to avoid unnecessary interventionKim Collins MEng, CEng, MIStructE, Conservation Accredited Engineer and Associate, Integral Engineering Design, Bath, UK
Margaret Cooke BSc, CEng, MIStructE, Conservation Accredited Engineer and Director, Integral Engineering Design, Bath, UK
IntroductionBath is a city in southwest England, famous for its creamy limestone, 18th-century architecture and hot springs. It is the only destination in the UK to have the whole city designated a World Heritage site. The River Avon runs through the centre of the city. Over time the general street level has become gradually higher as each layer of civilisation built on top of the last. On the west side of the river – the major historic centre of the city – street level now lies some 8m above river level.
This case study concerns Grand Parade, the street running along the river just south of the world famous Pulteney Bridge. Grand Parade is subject to heavy pedestrian and vehicular traffi c, particularly during the tourist season. Pedestrians and vehicles are protected from falling into the river by a masonry balustrade approx. 1.3m high, which is listed Grade II (Figure 1). Tourists lean on the balustrade in order to see Pulteney Bridge and its weir below (Figure 2), as do
W Figure 1Grand Parade
balustrade
E Figure 2Grand Parade
balustrade with Pulteney Bridge and weir
Synopsis
Evidence suggests that modern loading codes often signifi cantly overestimate real loads and that modern calculations may not be an appropriate way to assess the capacity of historic structures. This article describes how in situ load testing of the Grand Parade balustrade, Bath, was performed to confi rm its actual capacity and avoid the need for an intrusive strengthening scheme proposed by the local highways department which would have caused signifi cant damage to the historic fabric.
spectators at special events, e.g. fi reworks displays held in the Recreation Ground on the other side of the river.
The balustrade was beginning to show signs of degradation and concern was raised within the council as to whether it was fi t for purpose. The local highways department carried out an initial analysis of the balustrade and concluded that if one applied modern loads (pedestrian or traffi c) to the historic structure, then it would fail.
From this analysis the highways department suggested a strengthening regime which might be required to bring the structure up to modern standards. However, in consultation with the conservation offi cer, the team agreed that the regime of strengthening represented an unacceptable degree of intervention for a listed structure. Integral Engineering Design was therefore asked to provide an alternative risk-based approach to the problem with the aim of providing a more sensitive solution.
The stone balustrade is constructed with a solid stone plinth at the base on which sit stone pilasters at approximately 330mm centres; coping stones span between the pilasters (Figure 4). The historic balustrade drawings appear to show iron pins between the plinth and the pilasters and between the pilasters and the coping stones (Figures 5 and 6). The balustrade is split into bays approx. 3.0m wide, with a solid stone panel between each bay (Figure 7).
Approach to designOur fi rst response was to ask whether there was in fact a problem to be addressed. It was clear that the balustrades have withstood pedestrian loads for 100 years or more, so have eff ectively been load tested. Why was there a sudden need to justify the balustrade at this point in its history?
The answer to that question was twofold:
• the condition of the masonry was deteriorating, so both individual balusters
HistoryThe balustrade was constructed in three phases: the fi rst phase adjacent to Parade Gardens at the turn of the 19th century; the second phase from the turn in the road to the southern edge of Boatstall Lane in 1933; and the third phase from Boatstall Lane to Pulteney Bridge in 1934 (Figure 3).
"If one applied modern loads (pedestrian or traffi c) to the historic structure, then it would fail"
and the stooling at the base of the masonry were starting to spall and erode• as a society we are becoming more risk averse and certainly Bath and North East Somerset Council felt that it had a duty to consider the issue more scientifi cally – the approach of ‘it’s always been that way’ simply wasn’t acceptable
There were then two types of loads to consider: vehicle loads and pedestrian loads.
It was evident that there was no way in which the balustrade was going to be able to withstand vehicle impact loading. Discussion with the highways department led to an agreement not to consider vehicle loading. This was largely because the layout of the roads locally meant that it was highly unlikely that a vehicle would run into the barrier as a direct impact. The road runs parallel to the balustrade and there is no point at which a vehicle would approach the balustrade head on. If this had been the case, then the answer
would probably have been to place bollards or similar barriers at the kerb rather than try to strengthen the stone balustrade.
Horizontal loading from pedestrians leaning onto the balustrade was therefore agreed to be the critical load to consider in this case.
CalculationsIn order to assess the suitability of the existing balustrade, it was fi rst decided to do a simple two-dimensional (2D) analysis to check its resistance to overturning and horizontal (shear) loading based on the vertical dead load of the stone balustrade alone. This initial analysis ignored the solid panels and did not consider the behaviour of the structure in three dimensions (3D).
Overturning is the governing failure criterion and it was found that the existing balustrade is theoretically adequate to resist a characteristic load of 0.4kN/m applied at a height of 1.1m above ground level, well below the 1.5kN/m required by British Standards1.
The balustrade has slightly better resistance in shear and is capable of sustaining a load of 1.3kN/m (it was assumed in this calculation that shear failure would occur at the base of the balustrade where, from the historic section, it does not appear that the stones are tied together).
From the above calculations, it is obvious that the balustrade cannot resist the required loading in this way (Figure 8). A more detailed analysis was subsequently carried out considering arching between the solid panels located at approx. 3.5m centres. It was found that the dead weight of these larger solid panels was still insuffi cient by a factor of two to resist the reaction from the balustrade arching in this way.
It is impossible with 2D hand calculations to take into account all second-order eff ects which may be allowing the balustrade to work as existing. At this stage, one approach would be to attempt to model the structure in 3D using a fi nite-element package or similar. However, due to the complex shape of the structure and nature of its construction, it was felt that this was not the right approach to take, as the task would involve a signifi cant amount of time and might not yield a diff erent result.
Past experience of historic structures shows that they can often support signifi cantly more load than can be proved by calculation alone. Since the implications to the historic fabric of strengthening the balustrade to meet modern standards could be severe, it was decided to carry out a series of in situ load tests to confi rm the minimum failure load and failure mechanism. The testing would also confi rm where strengthening should be focused if it was still found to be necessary.
Derivation of testing loadModern loading codes often overestimate the actual loads that are likely to be encountered, e.g. the British Standards loading code (Eurocode 1)1 states that offi ces should be designed for 2.5kN/m2, but previous research2 has shown that most offi ces only ever achieve a load between 0.4 and 1.8kN/m2. As it is likely that the same conservative approach has been applied to horizontal loads over time, and given the age and importance of the structure, it was decided that it would be appropriate to carry
out further investigation before accepting the codifi ed loads.
The University of Bath has carried out some research on the horizontal loads generated in a rugby scrum, i.e. by groups of people very deliberately applying a horizontal load in a concerted eff ort3. The research concluded that the peak force generated by a male international rugby team upon entering a scrum is approx. 1.4kN,
with a sustained load of 1.0kN. These forces were generated by the ‘pack’ hitting the scrummage machine, as opposed to a single individual.
It was concluded after discussion with the highways team that that this would be a reasonable testing load, as it is unlikely that a greater load could be applied by passing pedestrians or even a crowd standing adjacent to the balustrade while watching the annual fi reworks display. A sustained testing load of 1kN/m was therefore assumed. In reality this load is likely be conservative, as the scrum height is at hip level (bent and angled legs) not at 1.1m height (Figure 9).
A factor of safety of 1.2 was decided on. This is less than the 1.5 given in EC1. However, The Institution of Structural Engineers’ guide to Appraisal of existing
structures4 defi nes the load factors for dead and imposed loads as a product of three factors: the load variation factor, the combination and sensitivity factor, and the structural performance factor. As the structure was to be tested to the maximum
"The peak force generated by a male international rugby team upon entering a scrum is approx. 1.4kN"
case the fi rst three tests yielded ambiguous results. Following a meeting with the highways department on site, it was decided that one of the locations was impractical from a traffi c management point of view. Five reaction frames were therefore set up for testing (Figure 10).
The load testing was carried out by Sandberg Consulting Engineers on 2 October 2014. The load was applied with a 900mm long loading beam perpendicular to the coping stones using a hydraulic ram (Figure 11). The load was applied in 0.3kN increments and a strain gauge was used to measure the resulting defl ections just beneath the loading beam (Figure 12). The full load was then removed and the cycle repeated a further two times.
Project focus
load that it was felt could conceivably be applied, with anticipated future testing at regular intervals, it was deemed acceptable to reduce the load variation factor. Given the accuracy of measurement, and taking into account second-order eff ects (i.e. full-scale testing), it was also felt that the structural performance factor could be slightly reduced.
TestingOne section of balustrade (zone between two solid panels) was selected from each phase of construction for testing. In order to be representative, average sections were chosen, i.e. not the worst case, but not the most perfect example either. Three further sections were also selected as ‘back up’ in
No movement of the balustrade was recorded during the fi rst cycle of loading of the fi rst section of balustrade; therefore, following discussion with Sandberg, it was agreed to increase the applied load from 1.1kN (1.2kN/m applied over 0.9m) to 1.5kN (1.67kN/m applied over 0.9m) (Figure 13). During the three cycles of loading of the fi rst section, we saw negligible defl ections in the order of 0.06mm and the balustrade fully recovered (behaving elastically) between cycles.
The largest movement encountered during the fi ve tests was in the order of 0.2mm and all but 0.02mm was recovered following the three loading cycles. This is well below the span/180 defl ection of a cantilevering structure we would commonly allow.
May 2015 Load testing a historic structure
Load applied at this height
N Figure 8Shear and overturning failure N Figure 9
Testing load
N Figure 10Scaff old reaction frame with loading ram in place N Figure 11
ConclusionsAll fi ve sections of balustrade subject to testing were capable of resisting the applied loading with negligible defl ections and no signs of impending failure. An acceptable defl ection for a 1.1m high balustrade would be around 6mm (span/180); the maximum defl ection achieved under the implied load was 0.2mm. The testing proved that the balustrade is capable of resisting more than four times the anticipated failure load through calculation alone. The elastic behaviour of the balustrade was unexpected as masonry is a non-ductile material and a brittle failure would have been expected.
The load was applied over a 900mm long beam to the centre of each bay. It is likely that the balustrade is resisting the load through an arching mechanism and the resulting load spread. Should the load have been applied along the length of the balustrade, the same conclusion might not have been reached. When initially assessing the test load, it was deemed that it would be unlikely that the balustrade would be subject to 1.67kN/m (1.5kN/0.9m) along its entire length and that this load represents a localised peak force which may occur at only one or two locations on any occasion.
The testing outlined in this article considered an overall failure of the structure caused by people leaning at coping stone level only. There remains the possibility that individual pilasters may need extra reinforcement in order to prevent local shear failure. The balustrade
E 1 British Standards Institution (2002) BS EN 1991-1-1:2002 Eurocode 1.
Actions on structures. General actions. Densities, self-weight, imposed
loads for buildings, London, UK: BSIE 2 Fitzpatrick A., Johnson R., Mathys J. and Taylor A. (1992) An Assessment
of the Imposed Loading Needs for Current Commercial Offi ce Buildings in
Great Britain, London, UK: StanhopeE 3 The Institution of Structural Engineers (1996) Appraisal of existing
structures (2nd ed.), London, UK: The Institution of Structural EngineersE 4 Preatoni E., Stokes K., England M. and Trewartha G. (2012) ‘Forces generated in rugby union machine scrummaging at various playing levels’, IRCOBI Conference Proceedings, Dublin, Ireland, 12–14 September, pp. 369–378
References
has suff ered from neglect over recent years and is showing signs of signifi cant weathering in some locations. It will now be subject to a maintenance contract,
all joists will be fully pointed and, where required, loose or corroded stones will be pinned in position or replaced.
Testing the historic balustrade in this way has allowed signifi cant intervention and loss of historic fabric to be avoided. It has been successful in proving that structures, particularly those designed before modern design codes were implemented, often have far greater capacity than anticipated and careful thought should be given to how these structures are analysed and repair works specifi ed.
AcknowledgementsClient: Bath and North East Somerset CouncilEngineer: Integral Engineering Design LtdLoad testing: Sandberg Consulting Engineers
"Testing… has allowed signifi cant intervention and loss of historic fabric to be avoided"
N Figure 12Strain gauge registering displacement of 0.02mm N Figure 13
