Characterisation of the tensile performance of bonding agents for the restoration of heritage dimension stone from southeast Australia

Abstract Through this research, we measure the direct tensile strength of three important southeast Australian heritage dimension stones and compare the results with published values. The three stones are Sydney Sandstone, Victorian Bluestone and South Australian Black Granite. This information is crucial for stone masons when they are selecting and cutting stone for construction projects, as they need to ensure that the stone they use and the repair will be strong enough to withstand the loads and stresses it will be subjected to in its final location. Results show that direct methods of testing provide tensile strengths that are 30–63% of the indirect methods. When the direct tension to compressive strength ratio is considered, results of 2–7% are calculated for these stones. This is significantly less than the typically assumed 10% for intact rock. Three masonry bonding agents have been selected to determine their capacity to re-bond the dimension stone. Highly variable results were achieved, with re-bonded tensile strengths ranging from 6% to 104% of the intact strength. This variability is most likely related to the heterogeneity of the stones and the surface preparations. Based on these outcomes, it is recommended that stone masons carefully consider the product that is used for restoration works and that test samples be prepared to confirm capacities. KEY POINTS Direct tension testing has been completed on Sydney Sandstone, Victorian Bluestone and South Australian Black Granite dimension stones. Direct tension results have not been previously published for these stones. Direct methods of testing provide tensile strengths that are 30–63% of the indirect methods on these stone types. When the direct tension to unconfined compressive strength ratio is considered, results of 2–7% are returned for these stones. This is significantly less than the typically assumed 10%. The performance of bonding agents on each of the stones provides variable results suggesting that careful selection of products is required and, if possible, test samples be prepared to confirm bond capacities.


Introduction
Dimension stone has been used as a building material for thousands of years, owing to its strength and durability, and has been a primary building material for structures associated with status, power and religion (e.g. pyramids of Giza, Machu Pichu, Moai on Easter Island, Stone Henge in the UK) (Patil et al., 2021). 'Modern' historical buildings constructed during the Australian Gold Rush and Federation Eras remain significant architectural pieces in the cities and towns around southeastern Australia (Victoria, New South Wales and South Australia). In each of these locations, historical government buildings and monuments are typified by their stone masonry. These buildings were constructed from local quarried dimension stone and remain a benchmark for cultural and societal development as presented in Figure 1 (Berrocal-Olave et al., 2021).

Damage to heritage stone buildings
Given the age of many of these buildings that date back to (or precede) Federation in 1901, there is a need to conserve and restore them from the effects of climate and urbanisation of our cities (Franklin et al., 2014). Usual damage observed associated with these buildings that requires rehabilitation includes discolouration, weather erosion, cracking and chipping. Typical damage profiles are presented in Figure 2. Cracking can be induced by weathering or subsidence related to underlying and adjacent construction activities.

Guiding principles for heritage restoration
Each state in Australia has policies or charter detailing the work necessary for the classification and restoration work that can be performed upon such historical buildings. Those for southeast Australian states include: The Burra Charter (ICOMOS Australia, 1988aAustralia, , 1988b, Heritage Acts (NSW Government, 1977;Victorian Government, 2017) and Planning and Design Codes (Department of Environment and Natural Resources, 1997;SA Government, 2016). In each case, these policies outline that repair works should, as far as practically possible, aim to provide solutions that are visually similar to the original stone used.
However, the policies do not consider the production phases of building or monument restoration. While we have the building standard AS 3700:2018 'Masonry Structures' (Standards Australia, 2018), which provides the minimum standards for construction on new buildings, we have no standards or codes of practice for restoration works on stone structures. As such, restoration works at the present time are performed in compliance with the British, Scottish or American heritage requirements. This requires experienced structural and geotechnical engineers to assess the damage to stone buildings and determine the serviceability of stone elements.
The restoration of cracking, weathering and spalling damage in many cases requires the replacement of portions of the stone block and the bonding of two stone sections (original and new) in place. The tensile strength of the original stone and the bond strength generated with the new piece can significantly influence the quality of the repair. Through this research, we directly measure the tensile strength of three important southeast Australian heritage dimension stones and compare the results with published values. The three stones include Sydney Sandstone, Victorian Bluestone and South Australian Black Granite. Furthermore, the results of bond tests on each of these stones using three common and commercially available bonding agents are completed to guide heritage restoration works.

Stones of heritage importance
A significant body of work exists associated with establishing the source and historical importance of the dimension stone that has been used in each of our heritage buildings (e.g. Cooper, 2019;Cooper & Kramar, 2014;Cresswell, 2019;Lewis, 2019;Peck et al., 1992;Walter, 2018). For this study, three of the most iconic stones are chosen.
Sydney (or Hawkesbury) Sandstone, historically known as 'Yellowblock', forms the bedrock for much of Sydney. The stone is durable and was a significant building material from the late 1790s to the 1890s. The stone has an average grainsize of 0.3 mm and porosity of 16.1% (Falconer, 2010;Pells, 2004;Standard, 1964).  (Davies, 2021), Victoria (Wikipedia, 2022) and South Australia (Cooper, 2019). Victorian Bluestone is a volcanic basalt that originates in western Victoria and dates at 4.5 Ma. Historically the bluestone was quarried by convict labour and was sent back to England as ballast in ships. The stone can be seen in buildings around the port areas of London. The stone has an average grainsize of 0.2 mm and porosity of up to 18% (Trigg, 2017;Walter, 2018).
Adelaide Black Granite commonly also called 'Imperial Black Granite' or 'Austral Black Granite' is a black norite, which has been quarried since 1958, 100 km east of Adelaide. The stone has an average grainsize of 0.02-0.2 mm and porosity of <0.5% (Adelaide Marble Specialists, 2022;Cooper, 2019;Young, 1993).

Tensile strength test methods
The tensile strength of stone is important, since it determines the maximum amount of stress that the stone can withstand before it breaks or fractures. This information is crucial for stone masons when selecting and cutting stone for construction projects, as they need to ensure that the stone and the repair will be strong enough to withstand the loads it will be subjected to in its final location. For example, consider that the head of one of the gargoyles on St Paul's Cathedral in Melbourne needed to be replaced owing to damage at the location of the white line in Figure 3.
The restored head would need to be bonded to the existing body. In order to ensure the long-term stability of the restoration, the bond strength of the new piece to the old needs to exceed (or be similar to) the tensile strength of the rock or a plane of weakness is created that may result in future damage localisation.
The direct tensile strength of rock is seldom measured owing to the complexity in test specimen preparation and difficulty in testing procedure/s. As a result, there are numerous indirect methods that are applied that include: Brazilian (ISRM, 2007;ISRM, 2018) Tensile Splitting (ASTM International, 2008) Three Point Bending (ASTM International, 2015) Flexural (ASTM International, 2020) These indirect methods require the sample to be subjected to compressive forces that generate a tensile failure of the sample. Examples of direct and indirect test loading conditions are provided in Figure 4.
Previous studies have shown that indirect tension methods overestimate the strength of geomaterials (Johnson et al., 2015;Perras & Diederichs, 2014). They are also prone to inaccuracies owing to variation in the size and shape of the sample. As such, to accurately characterise the tensile strength of a geomaterial, a direct method is preferred. Direct methods for rock are outlined in ISRM Test Methods (ISRM, 2007); however these methods require the use of a bonding agent to connect the sample to the loading platens. The use of a bonding agent is problematic, since few have a capacity suitable for rock and/or their elasticity impacts the measured strain results. As such, many researchers have sought methods to modify existing direct (Alhussainy et al., 2019;Guo et al., 2022;Zhang & Lu, 2018) and indirect methods (Liao et al., 2020;Resan et al., 2020) with various complex specimen shapes and platen couplings.

Published tension values
For each of the heritage stones considered herein, average stone property characteristics are available. In some cases, minimum performance criteria are established for use in heritage construction/rehabilitation (e.g. Franklin et al., 2014;Mann & Milevski, 2016). However, the properties are primarily related to the use of these stones as pavers or, for Sydney Sandstone, in large excavations for underground space (Keneti et al., 2021). Limited published data exist regarding the tensile strength of dimension stone that is useful to masons for heritage restoration purposes.

Direct tension sample preparation
For simplicity, owing to the complications in testing, a direct tension test for rock has been modified based on the test method applied to steel (ASTM International, 2021). This geometry is chosen, since it localises the area of maximum stress in the narrow portion of the sample, thereby providing a reliable tensile failure mode and failure surface for stress calculations. For this direct tension laboratory investigation, specimen geometries have been prepared in general accordance with ASTM370-20. However, the sample shape has been modified to a cylinder rather than a flat geometry. The final sample geometry is presented in

Test results
Seven to 10 samples of each rock type were successfully tested under direct tension. A tensile strain rate of 0.02 mm/minute was applied. The results are tabulated in Table 2. Typical fracture profiles of the samples are provided in Figure 7. An example is also provided of an unacceptable break that has not been recorded above. A summary of the results is presented in Figure 8 along with published indirect values for comparison.

Discussion of direct tension results
Based on the results of the direct tension tests (Table 3), in each case the indirect methods provide values that are higher than those measured. The Brazilian indirect tensile strength (ITS) of the Sydney Sandstone is the 'closest' to the direct tensile strength (DTS) with a DTS:ITS ratio of 0.63 (63%). This is in reasonable agreement with previous   (2022) investigations of these ratios for rocks by Perras and Diederichs (2014) that provide a ratio of 0.7 (70%) and 0.86 (Packulak et al., 2022). The flexural strengths of the bluestone and granite are consistent with ratios that have previously been measured by the authors on cemented paste fill at 0.45 (45%). These results are likely reflective of the heterogeneous nature of these rocks.
When the results of the measured and inferred tensile strengths are compared with published Unconfined  Compression Strength (UCS) responses for each of the stones, the results summarised in Table 4 are even more interesting.
The 10% rule of thumb (Hoek, 1966) for Tension:UCS is true when you are predicting an ITS Response, but we know these overestimate the strength: "This overestimation is hypothesised to occur owing to the confinement of the specimen based on the test geometry" (Packulak et al., 2022). The true ratio of tensile strength to UCS may provide an approximate estimate of tensile strength obtained indirectly. However, the actual tensile strength is typically significantly less than that provided by this ratio. The  tensile strength could be 2-7% of the UCS. This is a significant outcome that relates to the capacity design of structures using these materials.

Consideration of the performance of bonding agents
A total of three commonly used masonry bonding agents have been selected to determine their capacity to re-bond dimension stone. The three bonding agents included Tenax V R , Akepox V R 5010 and Megabond V R epoxy.
Megabond: "is suitable for bonding stone and tile to all masonry surfaces including cement sheet, cement render, plasterboard, timber and is suitable for both interior and exterior applications". It is noted to have a bond strength of 2.7 MPa on concrete (Vivacity Engineering Pty Ltd., 2015). Tenax: "designed for the nearly seamless bonding of porcelain, sintered stone, glass, engineered stone, marble and quartz, and natural stone products". It is noted to have a bond strength greater than 20 MPa on quartz and granite (Tenax USA LLC, 2016).
Akepox: "is mainly used in the stone-working industry for the weather resistant bonding and gluing of natural stone (marble, granite) as well as artificial stone or building materials (terrazzo, concrete)". It is noted to have a bond strength of 40 MPa (GmbH A, 2023).
For each of the epoxies, direct tension samples were prepared as described above, but manually broken in the middle by cutting through them and then re-bonding. In each case, a 1 mm-thick epoxy layer was applied according to the manufacturer's directions. Once the bonding agent had cured according to the instructions, the samples were tested according to the direct tension procedure outlined above. The results of the re-bonded tension tests are provided in Table 5.
Examples of each of the bonded samples and their failures after tensioning are presented in Figure 9. A comparison of the bonded strength results in relation to the intact (unbonded) strength is presented in Table 6.
To consider the bonding agent to be appropriate for stone restoration, it is expected that the re-bonded strength should approximate the intact (unbonded) strength. In all cases, apart from Akepox used on the granite, the re-bonded samples have lower tensile strengths than the original samples. Based on the average results for each bonding agent, Tenax and Akepox may be suitable for the rock types tested, whereas Megabond may not be suitable for any of them.
Based on the results in Table 6, it can be concluded that: 1. The only bonding agent that provides an acceptable result for the tests conducted is the Akepox on the granite with a re-bond strength of 104% of the intact (unbonded) strength. In relation to the other stones, the Akepox provides variable re-bond strengths between 44% and 55% of the intact strength. These bond strengths generated are approximately 10% of the published capacity of the product (40 MPa). 2. Tenax provides the most consistent results for the stones tested; however, its re-bond strength averages only 63% of the intact strength. The generated bond strength is approximately 10% of the published  capacity of the product (i.e. 40 Mpa) in the cases tested. Its performance is similar to that of the Akepox.
3. Megabond provides re-bonded strengths less than or equal to 22% of the direct tensile strength. The product may therefore be unsuitable for restoration works associated with these dimension stones. It is noted that the tensile bond strength of Megabond is only 2.7 MPa, and so its lower performance is not a surprise. It was able to generate a bond strength of approximately 20% of its published capacity in the cases tested.
Bonding agents are highly dependent on the nature of their bond surfaces (e.g. cement, wood, plastic, glass) and roughness (Ribeiro et al., 2023). In the case of the dimension stones studies, if we consider the porosity, then the granite has the lowest porosity at 0.5%, and the sandstone and bluestone are similar at 16% and 18%, respectively. As such, in this case, it is can be concluded that Akepox may be best suited for bonding low-porosity/finer-grained stones and provides similar (average) results to Tenax on higher porosity and larger grainsizes. As a result of these tests, it is recommended that stone masons carefully consider the product that is   used for restoration works, since variable results are likely based on the geology, surface preparation and bonding agent chosen.

Conclusions
The tensile strength response of geomaterials is a critical parameter for characterising its mechanical response when stressed or strained. However, the direct measurement of this parameter is rarely completed owing to the complexities in sample preparation and seating. As such, indirect techniques such as Brazilian and Three-Point Bending Tests are generally used for tensile strength characterisation. The completion of direct tension tests on a range of quarry stone from southeast Australia has highlighted some issues associated with assuming the indirect tension measurements of geomaterials provide an accurate measurement of true (direct) tensile strength. In the cases presented herein, the direct methods provide tensile strengths that are 30-63% of the indirect methods. When the direct tension:UCS ratio is considered, results of 2-7% are returned for these stones. This is significantly less than the typically assumed 10%. Furthermore, the performance of different bonding agents has been investigated with highly variable results achieved. For the stones and bonding agents tested, Akepox provided the highest tensile strength when used to bond South Australian Granite. However, the response of the Tenax was similar for the other stone types. The variabilities in the results can be attributed to the bonding capacity of the products that ranged from 2.7 MPa to 40 MPa and the porosity of the bonding surface. As such, it is recommended that stone masons carefully consider the bonding products that are used and that test samples be prepared to confirm bond capacities.