Structural evaluation of historical masonry walls utilizing non-destructive techniques – Comprehensive analysis

1 Introduction

Historical masonry walls are particularly challenging, and specific types of structural elements are required to analyze quantitatively [1]. In addition, their abundant variety is one of the critical factors when it comes to non-destructive testing (NDT) applications. It should also be noted that a significant amount of such structures (by default) is in the cradle of European civilization – namely, the basin of the Mediterranean Sea. This area is strongly affected by the seismic activity – Slovenia, Croatia, Portugal, Italy, Greece, and Turkey [2]. Only in the current century and in Italy alone three major earthquakes with disastrous impact on historical masonry walls can be listed: L’Aquila 2009 [3], Emilia-Romagna 2012 [4], and Central Italy 2016 [5]. Hence, in these areas, additional variable comes from different states of analyzed elements – a wall can be in its original form, in a damaged state after an earthquake, or after strengthening. Additional difficulties in terms of structural assessment are due to the fact that all these scenarios can occur within the same building.

Also, it should be noted that the assessment of historical objects requires a holistic approach, which includes a detailed and thorough investigation [6]. In many cases, significant diagnostic limitations arise from restrictions imposed on testing methods. The restrictions are related to the invasiveness of these methods and might be imposed by local or international regulations [7,8,9]. Specifically, Roca [7] discussed the ICOMOS/ISCARSAH guidelines, which (among other recommendations) prioritize the rule of minimum interventions; in ICOMOS [8], the safeguard of the object’s integrity is pointed out, while in ICOMOS [9] it is underlined that any invasive investigation should be extensively justified. Such decisions relate to the historical and cultural value of the considered objects.

Additionally, the mathematical models for masonry walls are quite different from those for typical beams or frames analyzed in the case of structural steel or reinforced concrete [10]. In the case of numerical analysis, shell elements or solid elements are much more capable of reflecting the specificity of masonry; however, at the same time, they are significantly more difficult to handle [11,12]. Masonry is a complex material to describe mathematically; hence, in engineering practice, various mathematical models are needed for different loading scenarios – compression, in-plane shear, or out-of-plane behavior. Also, a distinction must be made between static and dynamic loading [13,14,15]. Specifically, Lagomarsino and Giovinazzi [13] proposed two models to deal with seismic hazard – one based on the vulnerability of the walls and the other based on their capacity curves. Penna et al. [14] presented a macroelement model for simulating the cyclic in-plane response of masonry walls, applicable both in nonlinear static and dynamic analysis. Ghezelbash et al. [15] proposed a damaging block-based model (BBM) working in quasi-static and dynamic simulations. It is a numerical approach based on 3D finite elements with a plastic-damage material formulation; the masonry units interact by means of a contact formulation (Figure 1).

Figure 1

Distribution of unreinforced masonry buildings (in millions) across Europe; adapted from Crowley et al. [16].

Given all the enumerated factors, the practical application of NDT in the case of historical masonry walls is a challenging task. It requires not only knowledge in the field of testing methods but also strong familiarity with the specificity of historical masonry walls and existing engineering tools for structural assessment [17]. This statement also determines the setup of this article. First, the features of historical masonry walls are described, then relevant NDT methods, indications in codes and standards, and, eventually, the assessment methods. Finally, a synthesis and discussion are conducted, which confront the described components with the needs and capabilities of engineering communities. Specifically, based on the gathered and processed data/conclusions available in the literature, the knowledge gap is indicated. Also, the discrepancies, inconsistencies, and lack of communication between the scientific, legislative, and engineering communities are demonstrated. These issues might lead, in turn, to possible ineffectiveness and mistakes in the structural evaluation/assessment process. In conclusion, a roadmap and potential actions are provided to improve the current situation.

It should be noted that the description and specificity of the NDT methods are limited to the minimum related to the purpose of this study. For details of particular methods, the provided literature should be referred to.

2 Historical masonry walls – features

In this section, to demonstrate the specificity of the analyzed structures, some of the features are highlighted. Historical masonry walls are a broad term encompassing a variety of structures built with different materials, techniques, in different periods, and across various geographical and cultural regions [18,19,20].

An obvious variable is the material of blocks (units) that are inevitable components of masonry. Masonry typically is made of bricks or stones. Furthermore, historical bricks can also be divided by the type of material: burnt clay bricks, clay bricks, sand lime bricks, and soil bricks. The variety of rocks that constitute stone blocks is even greater, along with greater discrepancies in their mechanical parameters. Namely, stone blocks can be made of relatively weak sandstone (uniaxial compressive strength starting with 6 MPa [21]) or high-strength granite or basalt (uniaxial compressive strength up to 400 MPa [22]). All the discussed materials that are used as blocks possess their own physical and mechanical properties, various porosity, and chemical composition [23,24]. In the case of rocks, orthotropy or anisotropy can occur as well. As a consequence, inter alia, they react differently to aging [25], moisture content [26], natural chemical agents [27], freeze–thawing process [28], cyclic loading [29], or long-term phenomena like creep [30]. Moreover, a wide variety of physical and mechanical properties requires the adoption of dedicated renovation or strengthening methods and materials [31,32,33].

Similar considerations in terms of material can be applied to mortars. Namely, in historical masonry, walls can be found with mortars of different compositions and qualities. Their ingredients and, in turn, physical and mechanical properties are a result of local resources and tradition by the time of the structure’s erection. For instance, Lezzerini et al. [34] demonstrated that mortar used in the bell tower of St. Nicholas Church (Pisa, Italy) was obtained from firing local carbonate rocks, cherty limestone, and secondarily Mt. Pisano marble. Other works dealing with the composition of historical mortars as a function of local resources are, for example, as given previously [35,36,37]. In the study of Yaseen et al. [35], Roman mortars in ancient Jerash were analyzed, and within this one city and one historical period, two different types of mortars were utilized (in terms of binders). In the study of Riccardi et al. [36], an eighteenth-century object in Milan and a sixteenth-century object in Pisa were analyzed. In both cases, mortars exhibited hydraulic-type reactions, and the authors underlined that due to differences in the composition of mortars, different analytical methods were required. Gleize et al. [37] characterized mortars from the eighteenth to the early twentieth century in the State of Santa Catarina (Brazil), and the results show that the dominant binder is hydrated lime from the burning of seashells.

Although a complete understanding of the structure, mechanical properties, and deterioration level of blocks and mortar has been gathered, this knowledge is still insufficient for the structural assessment procedure. It must be noted that masonry is a composite that consists of units and joints. Hence, its overall properties are a result of the properties of these two components. It means that the structural behavior of masonry walls would also depend on the geometry of blocks (shape, size, roughness, level of processing) and mortar joints (thickness, spatial distribution). Overall, these features could be referred to as texture. The texture of masonry walls is another crucial variable in determining the wall’s properties, durability, and mechanical response. The texture of masonry walls alone is a vast area of research and engineering. Level of processing, or in other words, workmanship, has a significant impact on load transfer, both in the case of in-plane and out-of-plane loads; furthermore, it strongly affects the integrity of the wall and its cross-section. Also, the quality of surface treatment can have a significant impact on the durability of masonry. For instance, the presence of untreated surfaces promotes the spreading of internal wetting and, in turn, exposes the wall to amplified freeze–thaw processes. Exemplary classification of units regarding workmanship (after Peulić [38]) is presented in Figure 2.

Figure 2

Classification of the units depending on the workmanship: (a) rubble stone, (b) hewn stone in the form of slate, (c) cyclopean stones, (d) squared rubble with one face not completed, (e) squared roughly tooled rubble, and (f) ashlar unit with finished edges and faces in smoothed view (after Peulić [38]).

The shape and level of processing applied to units determine the level of regularity in masonry walls, which in turn has a crucial role in the unit–joint interaction and stress distribution at their interface. Figure 3 demonstrates a sketch representing the distinction between regular and irregular walls.

Figure 3

Distinction between regular and irregular masonry texture (after Szabó et al. [39]).

Figure 4 shows the scale of variety in possible textures. This figure represents the typology of texture identified just in the agglomeration of the city of L’Aquila in Central Italy (after Rovero et al. [40]).

Figure 4

Example of masonry types in L’Aquila (Central Italy) (after Rovero et al. [40]).

So far in the description, none of the mentioned features have been related to the cross-section of the historical walls. However, it is a crucial parameter, which in the case of the modern walls has minimal significance (modern walls usually consist of one layer, and if there are more than one layer, they are always properly interlocked). Hence, another aspect specific to historic masonry walls is the presence of two or three layers in the cross-section (Figure 5). Such layers are often called leaves or wythes. Walls with two or three layers are called multi-layered or multi-leaf walls. In the case of three layers, the central one is called the core or nucleus, whereas the outer layers are called the cladding. This central layer has a significant impact on the overall behavior of the masonry wall, which stems from the specific structure of the core. Cores are often made of material of lower quality – cobbles, pebbles, irregular stones, crushed bricks, debris, and various types of “on-site rubbish.” Mortar is often of lower parameters as well – for example, with less binder and laid chaotically. Therefore, a significant percentage of voids and cavities are present in the cores. This inconsistency and discontinuities within a cross-section create an ambiguous state of stress, which hinders the structural understanding of such structures.

Figure 5

Exemplary three-layered masonry walls. Left: example from Palermo (Italy) – irregular stone cladding and very thin core consisting of small cobbles and pebbles (own source). In the center: brick masonry with regular cladding and core with high mortar ratio – Meldert-Lummen (Belgium). Right: uncoursed random rubble stone wall, stone cladding, and core made of cobbles in a matrix of mud mortar (Maharashtra, India) (source: own and previous studies [41,42]).

The variety of possible core-cladding configurations is also significant. Figure 8 depicts three examples of different three-layered masonry walls.

Another factor related to multilayered walls (both two-layered and three-layered) is the presence and quality of interlocking between the neighboring layers. Unlike modern multilayered brick walls, which feature a clear and repetitive pattern of headers, historical walls are characterized by interlocking elements (called through-stones or keystones) in a less organized manner. The question of layers’ interlocking is vital in the context of out-of-plane behavior during earthquakes and in the context of long-term behavior. In both these cases, the monolithic behavior of the entire cross-section is crucial for correct load transfer and, in turn, the safety of the walls.

There are numerous articles and reports demonstrating failure schemes in multilayered walls triggered by insufficient cross-sectional interlocking [43,44,45,46]. Most of the failures are related to seismic actions. However, deterioration of connections over time, along with geometric eccentricity, also may be destructive. Such a case is shown in Figure 6, a two-layered irregular stone wall.

Figure 6

Free-standing two-layered wall in rural Poland, damaged through the separation of layers (source: own).

The issues of weak core and interaction of layers have been investigated intensely over the last three decades – see, for example, previous studies [47,48,49,50,51]. Precisely, in the study of Binda et al. [47], using the laboratory shear test (similar to the one shown in Figure 7), the difference in quantitative and qualitative behaviors of masonry without and with keystones is depicted. Laboratory testing of three-layered masonry specimens was also done in the study of Valluzzi et al. [48]. In this case, three types of strengthening were tested – injections, bed joints repointing, and transverse tying. The injections turned out to be the most effective; however, the authors underlined that each case (each specific wall) might yield a different response. Ferreira et al. [49] considered out-of-plane laboratory testing with varying levels of axial precompression. In conclusion, the importance of the presence and spacing of keystones in mobilizing the monolithic behavior of specimens was strongly underlined. Three-leaf wallettes constructed with mud mortar were tested in compression in the study of Meimaroglou and Mouzakis [50]. The obtained global failure mechanisms were in line with air lime mortars. In contrast, the compressive strength and the modulus of elasticity were, respectively, higher and lower than the results for air lime mortars. Numerical parametric analysis (based on experimental data) was conducted by Boscato et al. [51]. As variables were treated, the mechanical parameters of core and cladding, the proportions of specimens, and the strength of core–cladding connections were considered. The most important finding revealed that core–cladding interactions govern load transfer and capacity.

Figure 7

Shear test arrangement for multilayered wall (source: Corradi et al. [52]).

However, further research is still needed, as quantifying these phenomena is a challenging task that requires numerous specimens, real-scale testing (mostly on recreated models in laboratories – Figure 7), and considering several variables.

Given both the described specificity of the discussed structural elements and the strict requirements related to heritage preservation, the role of NDT in diagnostics and structural assessment of historical masonry walls is crucial. We next discuss the relevant NDT methods.

3 Testing

Given all the specific features described above, it is clear that structural assessment (both in terms of statics and dynamics) of historical masonry walls is a complex, demanding, and multidisciplinary task. One of the steps is to acquire the parameters of the structure, not only in terms of quality but also in terms of quantity. The possible place and role of NDT in the entire process are demonstrated in Figure 8.

Figure 8

Possible role of NDT in assessment procedure of historical masonry walls (after Binda et al. [53]).

Regarding NDT methods for historical masonry walls, they can be analyzed within different frameworks. Namely, the tests could be considered in terms of the blocks, mortar, or the compound block–mortar. Furthermore, the variety of textures and the number of layers (wythes) in the cross-section might be investigated. Information can also be categorized based on static or dynamic analysis, or the long-term behavior of the structure.

The division below is purposely structured in terms of specific masonry parameters and features. However, it should be noted that, in the case of historical masonry walls, besides the methods listed below, two universal methods should always be considered in the first step. The first one is visual inspection, and the second is the study of available archive documentation and historical resources [54].

3.1 Materials

3.1.1 Mortar – strength

The most challenging aspect of NDT in situ testing of mortar (namely, without extraction) is its small size and parameters varying across joint depths. This variability primarily relates to weathering processes and undocumented renovative actions over the centuries. Given these stipulations, choices are limited.

The use of a penetrometer (even though it could be classified as a slightly destructive method) seems to be the most suitable tool for quantitative outcomes. It produces results that can lead to an estimation of mortar’s compressive strength, obviously, along with the correlation curves. In favor of the method is its applicability to low-strength mortar, which, in the case of historic objects, is critical. Colla [55] demonstrated the application of the penetrometer to a real historic structure with practical outcomes for both engineers and architects. However, significant comparative campaigns – namely, projects that include both destructive testing and penetrometric systems are required to obtain more reliable data. To date, empirical correlations for historic mortars are still limited [56]. Schmidt Hammer type could provide similar data with a comparable level of confidence with a penetrometer; however, given its high energy impact, the application for weaker mortars (the case of historical ones) is strongly limited [57]. Besides, while a penetrometer reaches up to several centimeters in depth, a Schmidt or pendulum rebound hammer can give information only about the superficial part of the material, around 3 cm [58]. The latter method, although simple in application and non-invasive, is very highly susceptible to surface morphology variations and yields significant inconsistencies [57]. The last method with potential applications in determining compressive strength of mortars is the scratch test; however, the attempts described in the literature show that results considering only the extremely superficial layers of the joint are challenging to interpret and unsuitable for irregular masonry [59].

Nevertheless, it should always be noted that this kind of method, based on energetic/dynamic approaches, is highly influenced by the boundary conditions – for example, the stiffness of the units in masonry. Hence, mortar testing in walls composed of soft rocks might render different values, which in turn require abundant data and an experienced user. For example, Žalský et al. [60] demonstrated such an effect in the case of the penetrometer using steel and brick molds (Figure 9).

Figure 9

Compressive strength values influenced by the mold material – steel (blue) and bricks (orange) (after Žalský et al. [60]).

Active thermography might be used for determining the thickness or delamination of pointing mortar [61]. However, such information is only presumptive, and additional techniques or intrusive actions might be required.

Some promising results have been obtained in the field of crack detection in the propagation of mortar joints with the joint application of digital image correlation (DIC), acoustic emission (AE), and ultrasonic pulse velocity (UPV) [62]. The results of the experimental campaign regarding the so-called brick masonry triplets are shown in Figure 10. Among the most critical findings crucial for potential engineering applications was the fact that AE and DIC indicated the failure mode and damaged zone before visible cracking was present. However, it must be pointed out that it was a laboratory campaign, and the transition of this approach on-site would require overcoming several challenges, for example, noise from other parts of the wall.

Figure 10

Crack evolution on a triplet, various representations. (a) DIC strain; (b) AE events – planar orthotropic; and (c) AE events – planar isotropic localization (after Livitsanos et al. [62]).

3.1.2 Blocks – strength

One of the most common NDT methods for blocks is the use of the rebound hammer (both Leeb/Equotip and Schmidt hardness tests). The use of these tools for blocks in masonry is an extrapolation based on the experience with concrete [63]. Hence, similar limitations could apply: size of the specimen, moisture content, smoothness of the surface, and calibration effect. Especially the last factor is significant as an extensive range of material can be tested – several types of bricks (of various ingredients and workmanship) and numerous types of rocks (as already indicated in the study of Kržan et al. [64]). Some of the issues with this method are statistically analyzed in the case of the vintage clay bricks in the study of Borosnyoi-Crawley [65]. When considering this group of methods, it is essential to note that, in both mortar and block cases, the absence of extensive calibration means the results should be regarded as qualitative or semi-quantitative at best.

Given the uncertainties of hardness tests, the so-called SonReb method was devised – first for concrete – Cristofaro et al. [66] described its evolution. This method combines the results obtained from the Schmidt test and ultrasonic pulse velocity test (UPVT). Such an approach enables the reduction of the uncertainties and limitations related to the separate methods. The SonReb application for masonry blocks is currently minimal. An experimental campaign (along with destructive testing) was conducted by Cabané et al. [67], where they tested 20 types of bricks. Although the combined approach yields better correlation with destructive tests, a calibration or reliable database is still required.

In the study of Gomez-Heras et al. [68], a combination of UPVT with the Equotip test (instead of the Schmidt test) was investigated experimentally. The study is especially valuable as it dealt with 29 different types of rocks. The combined approach was confronted with available data from destructive tests. One of the main findings is a significant improvement in accuracy in the case of non-porous polymineral rocks.

The combination of Equotip and Schmidt tests was investigated for 11 rock materials typical for masonry walls [69]. However, for selected materials, this approach provided a modest improvement in terms of prediction accuracy.

All these methods require the user’s focus and experience, as they are highly susceptible to numerous factors.

3.2 Compound

Given that masonry is a composite of blocks and joints, a reasonable approach is to test the entire wall (compound) instead of describing the ingredients separately. Besides, having independent results both for the components and the compounds provides databases for double-checking and building correlation curves/equations. Hence, significant attention in both academic and engineering communities is paid to methods that give the resultant parameters of the masonry wall.

In the case of NDT methods, the compressive strength of the entire element is derived mainly through its correlation with its stiffness [70]. Hence, for this approach, the database of destructive data is inevitable.

Investigation of stiffness promotes different kinds of acoustic wave methods. These methods are constituted by a broad group of techniques relying on the relationship between the sound energy traveling in the material and various mechanical properties of this material. In general, this is achieved through the transmission and reception of mechanical (acoustic) waves of different frequencies. The waves are applied to a given medium – for instance, masonry. In case of any discontinuities (such as a crack) encountered by the wave, some fraction of the energy is reflected from the flaw surface. The reflected signal is processed into an electrical signal and represented on a screen. With knowledge of the wave velocity, the location can be inferred. Also, information like size, orientation, and location of the object/flaw, in some cases, can be acquired. Depending on the needs and goals in each situation, the type of wave might be chosen – longitudinal, transverse, or surface acoustic waves. In estimating stiffness and density, a correlation between these parameters and the wave velocity is utilized. Materials of higher rigidity and density allow sound to travel faster. A comprehensive description of these methods can be found in numerous textbooks [71,72,73].

One of the acoustic wave methods is the so-called indirect sonic impact method (ISIM). For the first time, this approach was adopted for stone masonry walls by Miranda et al. [74]. In this test, the wave velocities were measured along the direct and indirect travel paths of walls. Importantly, a reasonable correlation was obtained between stiffness from ISIM and stiffness from destructive tests. In the same campaign, the direct method and ultrasonic tests were applied as well. The first option also yielded a relatively good correlation, however, with a significantly higher number of tests compared to ISIM. The ultrasonic tests were found to be less reliable, mainly due to susceptibility to surface conditions (smoothness, roughness). The same authors carried out a similar campaign focusing only on ISIM [75]. The results were compared with the modulus from direct compression tests of 12 one-layered masonry panels, both with regular and irregular joints. ISIM correctly showed sensitivity to different wall types and good quantitative correlation with compression tests. However, it was still concluded that the method requires some cross-checking and should preferably be applied with other testing tools.

Spectral analysis of surface waves (SASW) was employed to diagnose stone masonry walls of Saint Justo and Pastor Church in Granada (Spain) [76]. Besides the elastic modulus, Poisson’s ratio was obtained. The agreement of both values was very good with the parameters obtained from destructive tests. However, for obvious reasons, the destructive tests were run on similar stones from the region instead of utilizing the structure itself.

There was one attempt at utilizing the UPVT method alone [77]. However, it was based on an extensive cross-validation procedure, which consisted of comparison with a laboratory compressive test. Hence, it has substantial limitations.

It is worth noting that all of the above methods are limited to homogeneous walls, optimally with only one layer of cross-section. Hence, their reliability decreases in the case of multi-layered walls. For example, in the study of Van Eldere et al. [78], sonic testing overestimated the stiffness of double-leaf masonry walls (Figure 11) by 23 times. The authors indicated that supplementary methods, like IE (Impact Echo), could provide information about the interlocking of layers and their thicknesses, for rectification of the results.

Figure 11

Specimens tested by direct and indirect sonic methods (Source: Van Eldere et al. [78]).

Another branch of methods is based on different types of dynamic analyses. The vibration tests can be either forced or natural. In the first case, the so-called experimental modal analysis (EMA) is carried out, while in the second case, operational modal analysis (OMA) is carried out. The advantage of OMA over EMA stems from the fact that forcing oscillation of large structures is much more difficult than utilizing natural vibrations, such as those caused by wind actions. Both methods provide natural frequencies, damping ratios, and mode shapes of the structures. Having such an output, numerical models can be calibrated. Part of the process is the adjustment of dynamic modulus (which can be converted to the static modulus). In this sense, vibration tests show potential for the indirect assessment of mechanical properties. A similar strategy was adopted by Tomaszewska et al. [79] for a historic bell tower and by Tomaszewska et al. [80] for a historic lighthouse.

3.3 Cross-section

As already demonstrated, the properties of cross-section and inner parts of historic masonry walls are significant in terms of their mechanical behavior.

In the case of cross-sectional diagnostics, the previously mentioned vibration tests are primarily used to detect local damage and cracks. For example, such an approach was applied to crack detection in the multilayered masonry wall of the bell tower (Figure 12 [81]).

Figure 12

OMA results, crack patterns: fronts (on the left), vertical cross-sections (on the right) (Source: Gentile et al. [81]).

Previously introduced wave acoustic methods are also used for cross-sectional investigation. Given their specificity, they are much more helpful for this task than for the assessment of mechanical parameters. Numerous experimental campaigns were conducted in this field. Sonic direct tests, sonic tomography (multiple pathways approach), and MASW (Multichannel Analysis of Surface Waves) were adopted by Valluzzi et al. [82]. The combination of methods allowed them to detect voids, cavities, different materials, and the effectiveness of grout injections executed on tested masonry panels. Qualitative information regarding the heterogeneity of masonry cross-sections was obtained through sonic methods in the case of the Cathedral of Noto. The research had a practical aspect, as it served to verify the state of damage and to indicate the possible renovation techniques in the reconstruction project (Figure 13) [83,84].

Figure 13

Exemplary specimen, tested cross-sections, and horizontal sonic tomography (Source: Valluzzi et al. [82]).

Impact-echo can also be used for the assessment of the cross-section. In the study of Sadri et al. [85], this technique was applied to study multilayered stone walls. The method was able to detect voids, cavities, the thickness of stones, and zones of high- and low-quality bonding between the units.

Ground penetrating method (ground penetrating radar [GPR]) – a geophysical technique based on the reflection of electromagnetic waves, is also utilized for diagnostics. A thorough review of existing experimental campaigns utilizing GPR was conducted by Binda et al. [83]. They concluded that this method has high efficiency for determining thickness, detecting metal intrusions, and identifying air voids. To some extent, it can be utilized for the study of historical redevelopment, delamination of facing layers, and moisture mapping of structures. Frąckiewicz et al. [87] presented a case study that employs this method for some of the listed purposes. However, it was indicated that some initial excavations were needed to calibrate the results; hence, this method also needs auxiliary semi-destructive actions. Similar conclusions were drawn by Palierak et al. [88], who found that GPR needed the support of boroscopy to be reliable.

A combined approach was adopted by Grzyb et al. [89] during the preliminary assessment of masonry walls in Malbork Castle. Specifically, they utilized both ultrasonic tomography and portable penetrating radar tests. Thus, they were able to identify voids inside the cracked wall, indicating, however, that the results were rather qualitative than quantitative. Furthermore, the capabilities of GPR as a function of different antennas were demonstrated and, if possible, confirmed with layer removal and drilling.

There is also a constant search for other methods, such as electrical resistivity tomography [86], which was used by Abu Zeid et al. [90] to assess the volume of grout injections into the core of the wall. Qualitatively, the results were promising; however, quantitatively, the method required significant calibration, which also excludes it as a standalone general diagnostic tool.

4 Standards and codes

Testing and obtaining mechanical parameters are only a part of the structural assessment process. In general, both designing and assessing procedures are described in standards and codes. Even though engineers are not always strictly bound to obey these documents, it is a common requirement of the investors to follow standardized procedures. However, in the case of historical masonry walls, the level of precision and detail provided in standards is much lower than in the case of new structures. Furthermore, given the fact that the current approach in dimensioning is to calculate as many features as possible, the lack of accurate guidance for historical masonry is frequently a source of confusion amongst the engineering community. A review of several standards and codes is given next.

In New Zealand, given that this country is strongly affected by earthquakes, historical masonry is analyzed mainly in terms of seismic actions. In 2017, a series of codes and standards were released to deal with this disaster [97]. For existing unreinforced masonry buildings, a volume was also issued [98]. In this document, a procedure for the assessment of entire unreinforced masonry buildings is provided (including interaction between walls and walls–floors or walls–roof interaction) (Figure 14). Some indications are also given for the quantitative description of masonry properties. However, most of them are based on destructive and semi-destructive techniques. The only non-destructive method (besides study of archives and visual observations) is GPR, which is advised for the determination of “the member thickness, metallic objects, voids, and other information.”

Figure 14

Assessment process of an unreinforced masonry building (Source: NZEE [98]).

The document also underlines the importance of the wall cross-section, indicating to identify any connections between wythes, determine the material of the core, and locate voids and cavities. It is stated that all these features contribute to determining the wall’s structural properties. However, no indications are given on how to quantify the contribution of these parameters (Figure 15).

Figure 15

Stone masonry cross sections in New Zealand. From left to right: dressed stone in outer leaves and rubble fill, stone facing and brickwork backing, stone facing and concrete core (Source: NZSEE [98]).

American Standard ASCE/SEI 41-13 [99], dedicated to existing structures, also provides some guidance for existing masonry. The document indicates that condition assessment shall include evaluation of the degradation level, deterioration of unit surface or mortar joint due to weathering induced by freeze–thaw cycles or frequent moisture saturation. Besides, the standard recognizes the importance of interlocking between wythes. The standard enlists the following non-destructive methods: UPV, mechanical pulse velocity, impact echo, radiography, and infrared tomography. A short description of each method is provided, along with its drawbacks and limitations, references to scientific articles are included. There are no indications regarding the quantitative interpretation of these tests and their integration into structural calculations. The Standard refers to other documents: FEMA 306 [100], FEMA 307 [101], and FEMA 308 [102]. First, besides the methods mentioned in ASCE, it also includes rebound hammer, SASW, and GPR. For each technique, a dedicated technical card is prepared containing the following: description, equipment, execution, personnel qualifications, reporting requirements, limitations, and references.

Canadian Standards CSA-S304.1-14 [103] does not refer to existing masonry structures; however, it gives some indications regarding the monolithic behavior of multi-wythe walls, which could be applied to historical masonry (after proper diagnostics).

Australian Standard AS 3700:2018 [104] also focuses on new structures; however, similarly to the Canadian code, some indications regarding interlocking between wythes are included.

European Standards, Eurocode 6 Design of Masonry Structures [105] does not contain information about existing masonry structures. Some indications regarding historical masonry walls might be found in part 3 of Eurocode 8 (Design of Structures for Earthquake Resistance) [106]. Appendix C of this part recommends the identification of the presence/quality of mortar, the presence of voids, and, in the case of multilayered walls, the identification of the presence, length, and spacing of through-stones. The code does not provide, however, any information about non-destructive methods or how to transfer obtained information into quantitative values for purposes of structural analysis.

Building Italian Code [107] and its accompanying Commentary [108] provide a set of mechanical values for various types of masonry in Italy. For each typology, compressive strength, shear strength, elastic modulus, shear modulus, and specific weight are provided. Besides, for all parameters, minimum and maximum values are provided (excluding specific weight). Furthermore, the code provides corrective factors for some specific features, like good-quality mortar or the presence of through-stones. This standard introduces three levels of knowledge that reflect the degree of diagnostics applied to the structure. The non-destructive tests (GPR, sonic tests, rebound hammer, thermography) are listed as allowable for the second level; however, no specific quantitative information is provided. The levels of knowledge also indicate the value of mechanical parameters – minimal, average, or probabilistic – to be adopted and the safety factors to be utilized.

In Italy, besides the approach given in the national standard, the so-called Masonry Quality Index (MQI) also functions – the first concept of this method was provided in the design code of the Umbria Region in central Italy [109]. Afterward, the approach was refined by Borri et al. [110]. This method might be based solely on visual inspections. The outcome is a set of mechanical parameters that is based on the total score of the masonry. Score in turn depends on the evaluation of the presence, partial presence, or absence of specific parameters that define the “rule of the art” in masonry construction. The reliability of the method was experimentally tested and calibrated on various existing wall panels by means of in situ destructive tests.

5 Structural assessment methods

After obtaining mechanical parameters, structural analysis can be conducted. At this point, several approaches can be adopted. Significant factors in this case are the loading scenario – static or dynamic (seismic actions) and the type of mechanical performance (in-plane or out-of-plane). Most analytical or semi-analytical methods consider the behavior of macro-panels. Geometry and mechanical properties of each panel are utilized to calculate values such as capacity, elastic, and plastic deformations. The state of stress, proportions, and boundary conditions of masonry panels determine the failure scheme [6,13,14]. In most cases, for the multilayered historical walls, some simplifications are adopted. For example, homogenized properties for the entire cross-section are assumed, or the strength of the weak core is neglected. Furthermore, if there is no reassurance about the effectiveness of connections between wythes, a separated mechanical performance is assumed [4]. However, features like voids, cavities, local degradation, and deterioration are difficult to include quantitatively.

A totally different set of tools is the numerical methods – finite element method, discrete method, distinct method, and mixed methods. A thorough review of each of these methods in the context of structural analysis of historical masonry is undertaken by Ghiassi et al. [111]. A different concept of classification was proposed by Daltri et al. [112]. Instead of differentiating by numerical methods, the classification was based on strategies in representing the masonry, specifically the interaction of joints and units. Four categories were proposed: BBM, continuum models, macro-element models (MM), and geometry-based models.

In general, with numerical methods and dedicated commercial software, many more features of historical masonry walls can be described mathematically. Namely, crushing in compression and cracking in tension, nonlinear stiffness, hardening, and softening. Furthermore, using contact elements, the limited strength of through-stones can be captured; with adequate modeling techniques, voids and cavities might be represented. A masonry can be described as a whole – macro-modeling or with separated joints and units – meso- or micro-modeling [113]. In the case of macro-modeling, the mechanical parameters are derived either directly from testing [114] or by using homogenization [115]. Homogenization could be subsequently divided into three groups: (a) a priori approach [116], (b) step-by-step multiscale approach [117], and (c) adaptive multiscale approach [118]. Frequently, the level of anisotropy and irregularity of historical masonry walls is so high that a direct continuum isotropic approach is the most effective and reasonable approach (Figure 16). Proper constitutive models enable modeling of creep [119]. Capture of various uncertainties and scatter of data is possible by using adequate probabilistic tools and procedures [120]. Hence, numerical models provide significantly more opportunities to utilize data obtained from NDT. The drawbacks of numerical methods include the high skills required from users (mostly professional engineers), the time-consuming preparation of models, limited access to specialized software, and possible convergence issues related to a high level of nonlinearity. Furthermore, even though undeniable progress has been made in terms of numerical modeling strategies for masonry structures, each computational approach exhibits distinctive shortcomings and can be reliably applied to a particular problem (for example, in-plane or out-of-plane loading).

Figure 16

Irregular and various textures of masonry (on the left), results representing the level of damage on a continuum isotropic numerical model (Source: Bartoli et al. [121]).

6 Overall discussion

This study covers and analyzes a specific area of structural engineering – precisely, structural evaluation/assessment of historical masonry walls with application of non-destructive methods. Such a defined problem is significant, particularly in the case of historical objects, where interactions with their tissue are restricted, and non-invasive tools are frequently the unique source of data gathering. The layout of the article is deliberately unorthodox. Particularly, four aspects are discussed: features of historical masonry walls, NDT, different building standards, and, finally, assessment methods. While discussing the specificity of historical masonry walls, it was underlined that, besides typical mechanical parameters – compressive strength and elastic modulus, historical walls are characterized by the presence of two or three layers (wythes). Frequently, the core of the wall is of a different structure than the cladding; the core may possess voids and cavities. The high variety in materials and texture was emphasized as well; high heterogeneity and possible weak interlocking between wythes were also indicated. Since the concept of practical outcomes drives the article, the NDT methods were associated with subsequent features of historical masonry walls. In other words, the discussion regarding NDT methods was governed by previously enlisted parameters of masonry. Hence, some of the methods could be recalled a few times. Given the fact that the literature describing the theory and procedures of NDT testing is relatively abundant, for the sake of brevity, these methods are usually not described. As the NDT methods are intended to serve common engineers in everyday practice, an overview of several standards and codes was provided. Particularly, it was analyzed whether the NDT methods are included in standards for the structural assessment of historical masonry walls. Then, the form, detail, and practicality of such inclusion were investigated. Special attention was paid to any quantitative indications, as assessments ultimately rely on numerical outcomes. Finally, the structural assessment methods were presented, both analytical and semi-analytical and numerical. Primarily, numerical methods were investigated as their role in engineering steadily grows; these methods are under constant development, and they also have higher potential to utilize data obtained from various non-destructive tests.

7 Conclusions

The comprehensive analysis of the stated problem led to the following conclusions and remarks.

• Choice and applicability of given NDT methods depend not only on the material considered but also on the type of structural elements and on the type of structural analysis that needs to be conducted. As a matter of fact, it is a general rule concerning any structure that was confirmed within this research.

• The driving factor of structural assessment in the case of historical masonry walls is their peculiar set of features. Namely, in addition to typical parameters regarding modern masonry walls, like compressive strength or stiffness of units and joints, a significant role is played by the geometry, texture, size, and shape of units; the number of wythes, the interlocking between them, and the quality of the central wythe (materials, voids, cavities). Based on this factor, NDT methods should be accordingly adopted. However, beforehand, some of the wall’s parameters are unknown. Hence, the choice of testing methods itself might be an iterative process.

• The choice of NDT methods is also strongly dependent on the purpose of the assessment. Namely, different parameters and consequently different testing methods would be needed in the case of in-plane loading schemes, out-of-plane loading schemes, or the evaluation of creep effects. There is always a possibility that all the purposes would have to be considered in parallel.

• The development of codes and standards should take into consideration the current state-of-the-art in the fields of numerical tools and NDT methods. The research demonstrated that even if standards list NDT methods (along with their description and purpose), there is no information regarding the quantitative application of the results and their incorporation into the structural assessment process.

• Besides, many codes, including Eurocode, do not directly address the assessment of existing structures. Also, many of them do not consider the application of NDT methods in the process of data acquisition.

• The already mentioned lack of quantitative indications for NDT in standards and codes comes from the fact that for these methods, such indications in general rarely exist. In most cases, to correctly interpret NDT output in terms of quantity, some minor-destructive or destructive method would be needed. For instance, sonic wave and GPR methods can identify voids, cavities, or loose materials; however, to convert these data into quantitative results, initial boroscopy or wall opening is necessary for verification and calibration.

• The closest to a stand-alone technique is the utilization of different types of energetic methods (hammer, pendulum, etc.) for the assessment of compressive strength. However, an extensive database is needed for interpretation; furthermore, results obtained for joints might depend on the stiffness of units and vice versa.

• Research indicated that one of the most suitable pathways to increase the role of NDT might be the creation of extensive databases and the elaboration of correlation equations and curves. However, to proceed with this concept along with NDT testing, destructive testing should be included (compressive tests, diagonal shear tests, creep tests, etc.). Obviously, given the cultural value of discussed structures, such attempts in situ on real structures are and would be very rare. Hence, laboratory tests on properly erected specimens (using desired historical materials and techniques) should be conducted. This strategy, on the other hand, requires thorough research programs, which in turn require ample financial resources, probably with collaboration at an international level.

• Most of the demonstrated methods require significant experience, access to correlation curves or datasets, and good judgment. As almost all cited authors indicated, the results are very sensitive and tricky to interpret. Such a situation may lead to erroneous conclusions and decisions in the case of users who are less conscious but are equipped with high-tech instrumentation. Hence, adequate legislative bodies and organizations should provide dedicated workshops, training, or guidelines.

• From the engineering perspective, it is essential to support the development and refinement of easy-to-apply tools like the MQI, which give engineers a clear path to proceed in the assessment process. Significantly, such methods might be based exclusively on archive documentation (if it exists) and visual inspections. The original MQI is developed based on Italian masonry walls; therefore, additional research is needed to tailor this method to different construction techniques in other parts of Europe and the World.

• It should be noted that the development of non-destructive tests should be carried out along with the development of calculation tools and assessment methods. Namely, large sets of specific data, even with a significant level of detail, become useless in the case of a lack of proper and reliable routines that transfer these data into practical applications.