Discussion of “Learning from Failure of a Long Curved Veneer Wall: Structural Analysis and Repair” by Paulo B. Lourenço and Pedro Medeiros

Hall, Christopher and Hoff, William D. and Hamilton, Andrea (2014) Discussion of “Learning from Failure of a Long Curved Veneer Wall: Structural Analysis and Repair” by Paulo B. Lourenço and Pedro Medeiros. [Review] (https://doi.org/10.1061/(ASCE)CF.1943-5509.0000491)

[thumbnail of Hall-etal-JPCF-2014-Discussion-of-Learning-from-Failure-of-a-Long-Curved-Veneer-Wall]
Text. Filename: Hall_etal_JPCF_2014_Discussion_of_Learning_from_Failure_of_a_Long_Curved_Veneer_Wall.pdf
Accepted Author Manuscript

Download (352kB)| Preview


The authors, in an interesting and valuable paper, describe the failure of the brick-masonry veneer façade of a multiuse public hall in Gondomar, Portugal. Damage occurred within 2 years of construction. The veneer was a single leaf of continuous brick masonry tied to a RC structural wall, forming a cavity of nominal width 0.07 m that was partly filled with foamed polyurethane. The veneer wall was 242 m in length and 15 m in height, without movement joints, and extended around most of the elliptical perimeter of the building with portions facing north, east, and south. From their site investigations and technical analysis, the authors attributed the failure primarily to effects of “the irreversible expansion of clay brick,” apparent both from cracking and from extensive out-of-plane deformation of the wall, which had widened the cavity to as much as 0.13 m. Effects were greater on parts of the wall facing south. The failure analysis made use of a power law proposed by Wilson et al. (2003) to describe how expansive strain develops in fired-clay ceramics with time. Here, the discussers comment on recent fundamental work on moisture expansion in brick, and in particular on its temperature dependence, matters of direct relevance to the paper under discussion. The discussers’ comments support and extend the conclusions of the authors, with which the discussers broadly agree. Irreversible moisture expansion occurs as a result of slow chemical reactions between components of the fired-clay ceramic and environmental moisture (Hamilton and Hall 2012). The magnitude of the expansion varies strongly with brick mineralogy and kiln firing history, but a predictive model for expansion based on these factors does not yet exist. However, in general, highly crystalline engineering ceramics produced at high kiln temperatures expand less than low-fired ceramics with a higher amorphous content. The penalty is that high-fired ceramics tend to be more brittle and prone to cracking. It is now established that the expansive reaction continues indefinitely, although at a diminishing rate over all timescales; therefore, there is no well-defined time at which it ceases. Recent reanalysis of published data (Hall et al. 2011; Hall and Hoff 2012) shows that the equation e=at1/4 accurately describes expansion strain e over periods of time t as long as 65 years. It follows from this equation that expansive strain at 16 years is double the value at 1 year and three times the 1-year value at 81 years. The persistence of the expansion reaction, albeit at a diminishing rate, emphasizes the need to incorporate appropriate movement joints in masonry design. The authors mention the possibility of using a “poor mortar” to accommodate some of the expansive strain. The use of weak mortars undoubtedly explains the absence of expansion damage in some much-older buildings with thick brick walls. However, the discussers consider that in thin brick veneers, such as those used in Gondomar, a weak mortar is potentially dangerous. It is unfortunate both for design and for failure analysis that the test procedures generally used to characterize clay brick do not provide values of the expansivity a that are needed to apply the equation e=at1/4 . Accelerated steam tests, such as EN772-19 cited by the authors, are at best semiquantitative. In our view, it is essential to determine the expansivity from measurements of expansion strain made over an appropriate period of time under controlled conditions (Hall and Hoff 2012). The discussers also draw attention to the important practical matter of the temperature dependence of the moisture expansivity (Hall et al. 2013). The fact that moisture expansion is the direct consequence of a chemical rehydroxylation reaction (Hamilton and Hall 2012) ensures that the expansivity increases notably with temperature. Available data indicate that the activation energy (which controls the temperature dependence) is about 70 kJ/mol. This means, for example, that the expansivity a of any brick material is about 60% greater at a temperature of 30°C than it is at 10°C. Thus, if a limit expansion strain (say, 1×10−3) is reached in a particular material in 50 years at 10°C, the same strain is attained in the same material in only 7 years at 30°C. It seems likely that its strong temperature dependence explains why moisture expansion is perceived differently in different geographical regions [e.g., McNeilly (1985)] and generally receives more attention in regions with warmer climates, such as Australia, southern Asia, and Brazil. However, in any particular region, the magnitude of expansion and the associated damage within individual buildings are influenced by local temperature variations, in particular variations due to solar heat gain. In the Gondomar structure, deformation and cavity expansion were greatest in parts of the structure with a southern aspect, where the summer temperatures of the veneer are highest. The influence of aspect here is presumably exacerbated by the open situation of the building and by the insulation of the cavity where large temperature gradients might be expected. The discussers believe this large gradient acting over a thin veneer may partly explain such dramatic damage over a short period of time. A thicker brick cladding would probably fare better. Undoubtedly, there are also seasonal modulations of the expansion. The discussers have shown elsewhere how related thermal effects in the rehydroxylation of archaeological ceramics may be calculated (Hall et al. 2013).