This report details the data verification process for the information utilized in the publica- tion ”Empirical fragility and ROC curves for masonry buildings subjected to settlements” [3]. The data are stored in the 4TU.ResearchData repository [4]. Data collection and anal- ysis took place between 2020 and 2021, with the final version being published in 2023. A thorough check was conducted before and after the publication of the manuscript to ensure data quality. While the first check was conducted by the Authors of [3] during the initial data collection, this report focuses on the second check, carried out after the pub- lication. In this document, the adopted methodology for the data verification is detailed and the results of the procedure are presented.

This report begins with Section 2, which outlines the methodology used for the data verification, whereas Section 3 summarises the results. Section 4 offers a discussion and conclusions. Finally, appendix A contains the calculations that support the verification checks.@en

In the pursuit of introducing bacteria-based self-healing mortar for masonry repair, this study examined the potential of incorporating a poly-lactic acid (PLA) agent—already established in concrete repair—into cement-lime mortars, typical of historical constructions. Testing prisms constructed with varying lime/cement ratios revealed decreased flexural and compressive strength in high-cement-concentration mortars upon the addition of the agent; for mortars with high lime concentration, however, the agent led to an increase in both strengths. Furthermore, the agent's potential to self-repair was confirmed by allowing the remaining portions of tested samples to heal under humid conditions. Irrespective of mortar composition, cracks were resealed thus confirming the aesthetic, and potentially watertightness, restoration.@en

Cracks represent a prevalent form of damage in masonry structures, posing not only aesthetic concerns but also compromising structural durability; therefore, they are undesirable and need to be repaired. The repointing technique is traditionally implemented in this context, especially in historical masonry. However, this method fails to provide a long-term solution, leaving structures vulnerable to future damage. The paper investigates the applicability of a bio-based self-healing mortar to enable autonomous repair of masonry. This innovative mortar, developed to repair concrete structures, was implemented to explore the capacity of couplets to recover their original bond capacity and aesthetic aspect after multiple damaging events. Specimens built with calciumsilicate and clay bricks were subjected to subsequent cracking cycles using a crack-mouth-opening -displacement controlled bond-wrench test. Experimental results showed that self-repair, in terms of bond restoration and aesthetic filling of cracks, occurs even after multiple cracking cycles when the bio-based mortar is used with both types of bricks, optimizing the autogenous healing (intrinsic) of cement-based mortars. The effectiveness varied also according to the types of brick and healing environment used, e.g. under humid conditions (RH ~ 95%), 50% vs 80% of the original capacity was regained in fully separated couplets made respectively with clay and calcium-silicate bricks.@en

To quantify damage to unreinforced masonry structures, sensitive to tensile stresses and thus to vibrations caused by induced earthquakes in the north of the Netherlands, a bottom-up or physical approach has been followed. First, the crack-based damage expected on these common structures, was defined on a continuous scale (Ψ) based on width, length and number of cracks, so that it could also be quantified. Then, the initiation of cracks was investigated on full-scale masonry walls, single-wythe and of fired-clay bricks, using high resolution Digital Image Correlation; at 0.1 mm in width, cracks become visible. Further, the propagation of cracks, as they widen and lengthen, was monitored during repeated and cyclic testing. At repeated, identical in-plane drift, cracks still propagate and damage increases, though the increase is minor.

The experiments on walls and spandrels, displaying horizontal, diagonal, and vertical cracks, were used to calibrate numerical, finite-element-method models. These reproduced the behaviour of the tests also in terms of crack patterns and propagation, besides stiffness, strength, and hysteresis. Then, the models were adapted to explore the effect of earthquake vibrations, also in combination with existing damage caused by settlement-like actions. In this manner, the effect of initial damage could be quantified. Several other parameters were varied, such as the material strength, the geometry of the masonry walls, the soil properties in a soil-structure interaction interface, the record, PGV, and repetition of the vibrations, and the intensity of initial damage.

Relationships between these parameters and damage were captured into a surrogate model, which was used in a MonteCarlo simulation to determine the probability of damage. The fragility curves reveal, for instance, that fired-clay brick walls with no visible pre-damage (Ψ0=0) have a 5% chance of visible damage (Ψ≥1) at a PGV of, for example, 10 mm/s, a probability that rose to 20% if the walls had undetectable pre-existing damage (Ψ0=0.5). The probability of exceeding aesthetic damage (Ψ≈2.5) at this PGV is less than 1%. A lower PGV is associated with a smaller probability of damage.

Furthermore, it was concluded that repeated events lead to an increase in damage of about 10% for five similar events. This increase is not noticeable. In a sequence of events, similar events accumulate little damage and increases appear when larger events are experienced by a masonry structure.@en

Drift limits are useful thresholds; during design or retrofitting analyses, engineers can compare the expected behaviour of a structure to drift limits that predict when the structure will reach a certain condition. This helps ensure that structures satisfy specified performance goals when exposed to certain hazards. Masonry walls are susceptible to damage from lateral in-plane actions such as wind or earthquake loading; ensuring that in-plane drift remains sufficiently small will help limit this damage. Drift limits based on crack-based damage are scarce, however, with DS1 limits being extrapolated from higher damage grades based on structural strength capacity or ductility. In this work, crack-based damage is evaluated on a multitude of full-scale experimental walls surveyed with digital image correlation. This method observes the initiation and propagation of cracking. Cyclically incremental in-plane tests provide a range of drift-damage relationships. These are explored with machine learning to determine influential predictors and ultimately establish drift limits for light damage. Two types of brick masonry are explored: fired-clay and calcium-silicate. For the latter, light damage begins at an in-plane drift of 0.5 mm/m and can extend to 4.8 mm/m (or 0.48%) for the former before the masonry surpasses light damage and reaches structural damage grades. In comparison to drift limits set by other authors and (international) guidelines to characterise light damage, significant damage, or the ultimate capacity of masonry walls, the resulting drift limits for light damage from this work are set directly on the basis of experiments and are in good agreement with other authors. Most importantly, all the consulted values for ultimate capacity are much larger than the upper threshold for light damage determined herein, with limits for significant damage in the same order of magnitude. This result verifies the accuracy of the experimental crack-based characterisation used to establish the drift thresholds.@en

The July 2021 flood heavily affected many inhabitants, buildings and critical infrastructure throughout Germany, Belgium and the Netherlands. Specifically, the Ahr Valley (Germany) showcased the destructive power associated with these extreme events. Hence, this region was the focus of a field survey, aiming at describing the flood-induced damage to buildings and assessing the possible underlying processes that led to structural failures. The field assessment revealed a close connection between building failures and (1) local flow depths and velocities, (2) building location, (3) distance from the riverbank and (4) construction type. Although it is difficult to identify the exact causes that induced failures, the detailed assessment revealed that damages mainly originated from local scour and hydraulic loads, often unevenly distributed around buildings. Importantly, many buildings were significantly affected by (large) floating debris impacts and damming, both responsible for additional loads, highlighting their importance in flood-resistant building design. Furthermore, data showed that buildings near the riverbanks and in the upstream part of villages were more severely damaged. Altogether, data provide a better understanding of the flood processes that lead to building failures, fostering future research towards the development of safer protection measures and more effective flood risk management strategies.

@en

In this study, 2D and 3D modelling strategies are used to represent the behaviour of historical masonry buildings on strip foundations undergoing settlements. The application focuses on a two-story building, typical of the Dutch architectural heritage. An improved 2D modelling is presented: It includes the effect of the lateral walls to replicate the response of the detailed 3D models. The masonry strip foundation is modelled and supported by a no-tension interface, which represents the soil-foundation interaction. Two settlement configurations, hogging and sagging, are applied to the models, and their intensity is characterized using their angular distortion. The improved 2D model that includes the stiffness and weight of the lateral walls agrees in terms of displacements, stress and damage with the detailed 3D models. Conversely, the simplified 2D façade models without lateral walls exhibit different cracking, and damage from 2 to 7 times lower at an applied angular distortion of 2‰ (1/500). The improved 2D model requires less computational and modelling burden, resulting in analyses from 9 to 40 times faster than the 3D models. The results prove the importance of identifying and including the 3D effects that affect the response of structures subjected to settlements.

@en

In the Netherlands, the potential damage to the building stock due to subsidence phenomena has recently received increased awareness. However, evaluating and predicting damage to buildings in subsiding areas is a complex task that requires associating the vulnerability of exposed structures with the intensity of the subsidence hazard. Considering the widespread presence of subsidence-related damage to the built heritage, the focus of this study is to provide empirical-based insights to assess and forecast subsidence damage to masonry buildings. A rich dataset with manual levelling measurements was collected comprising 386 surveyed masonry buildings, mainly low-rise (terraced) houses built before 1950. Of the total set of buildings, 122 cases rest on shallow foundations and 264 on piled foundations. For each building, the recorded damage is related to the settlement, calculated from the bed-joint levelling measurements, using four different intensity parameters, namely differential settlement, rotation, relative rotation and deflection ratio. These four parameters are appraised in their capacity to effectively predict the intensity of the damage. The Receiver Operating Characteristic (ROC) method is used to evaluate the relative efficacy of the selected hazard parameters. The rotation, the relative rotation (angular distortion) and the deflection ratio are observed as the most accurate when predicting the intensity of damage, while the differential settlement appears less accurate. Additionally, the dataset was used to generate empirical fragility curves where the probability of damage is described as a function of the aforementioned parameters. Thresholds were set to distinguish between the light damage and the functional and structural damage state. At a relative rotation of 1/500 masonry buildings on shallow foundations were observed to reach or exceed light damage with a probability of 13%, and functional and structural damage with 5%. The availability of the bed joint levelling measurements made it possible to classify eight recurrent settlement profiles, including both symmetric and asymmetric profiles, associated with both the overall deformation and the rigid rotations of the surveyed buildings.@en

This paper presents an improved 2D modelling strategy which aims to represent the behavior of historical unreinforced masonry buildings on shallow foundations subjected to ground settlements. The application is presented with reference to a two-storey building, typical of the Dutch built heritage. The novelty comprises the inclusion of the effect of the lateral house-to-house separation walls of such old buildings. Additionally, the masonry strip foundation is modelled and supported by a boundary interface representing the interaction between the soil and the foundation. Two realistic hogging and sagging settlement configurations are applied to the model and their intensity is characterized using the angular distortion of the settlement shape. The response in terms of damage and deformations of the proposed modelling strategy is compared with the ones of five selected approaches based on the state of the art. For all the selected models, the damage severity is quantified objectively by means of a scalar parameter, which is computed considering the cracks’ number, length, and width.

The results of the proposed 2D model agree in terms of displacements, crack patterns and damage with the 3D models. On the contrary, the façade models that do not include the effect of the lateral walls do not exhibit the same cracking and damage, resulting in lower damage and deformations for the same applied angular distortion. Accordingly, the proposed modelling strategy requires less modelling complexity and the analyses are 9 to 28 times faster to run with respect to the 3D models. The efficient and accurate model allows performing a wide number of analyses to easily investigate the role of the various building’s features.@en

Innovative solutions for seismic-retrofitting existing structures are currently required, as often traditional strategies are expensive, non-reversible, highly invasive, and/or fail to address both serviceability and ultimate limit states together. The present paper describes a preliminary experimental campaign performed at TU Delft to investigate an innovative structural glass window for strengthening masonry buildings. To this purpose, a prototype composed of a timber frame, a semi-rigid adhesive, and a 20 mm thick structural glazing layer was designed. The prototype aimed to improve the structure’s behavior against minor but more frequent service vibrations (SLS), as well as against ultimate ones (ULS). Specifically, an increase in the structure’s in-plane capacity and stiffness was targeted to reduce cracking at low drifts/displacements, while at larger drifts, the adhesive’s tearing and timber crushing were used to activate damping. To evaluate the prototype’s performance, a quasi-static, cyclic, in-plane test on a strengthened full-scale wall was performed and compared with available data on a similar, yet unstrengthened, wall. Although the benefits were not pronounced in terms of cracking and energy dissipation, the implementation of the proposed strategy provided an increase in terms of initial stiffness (18%), force capacity (8%, 36%), and ductility (220%, 135%). This outcome provides the ground for numerical studies that will help better delineate the proposed strategy and improve the current design.@en

This paper demonstrates the use of non-linear finite element modelling to investigate the response of structures subjected to different shapes of subsidence-related ground settlements. The approach is presented with reference to a two-storey unreinforced masonry façade resting on a shallow foundation. Eight realistic settlement shapes, based on field and literature data, are applied in the model with increasing intensity. The intensity of the subsidence profiles is characterized using their (angular) distortion. The extent of the induced damage on the façade is objectively and directly quantified by a damage parameter, based on the number of cracks, their length and opening. The performance of different settlement indicators and corresponding limiting values, typically employed in the state of the art, is in this paper discussed in relation to the damage modelling strategy; these are observed to be dependent on the shape of the settlement profiles. The aim of this paper is thus to provide insight into the extent to which the vulnerability of masonry buildings depends on the shape of the subsidence pattern and may serve as a warning not to use (deterministic) damage indicators such as angular distortion without considering the settlement shape.@en

Cracks are one of the most common expressions of damage in masonry structures. Aside from aesthetic issues, they can compromise the overall behaviour of the structure; therefore, they are undesirable and need to be repaired. The repointing technique is traditionally implemented in this context, especially in historical masonry. Nevertheless, future damage is not prevented and may arise again, thus requiring renewed repointing interventions. The paper describes a preliminary study conducted at Delft University of Technology to investigate the applicability of the innovative self-healing technology to enable an automatic repair of masonry cracks. A bacteria-based self-healing mortar, developed to repair existing concrete structures, was implemented to explore the capacity of couplets to recover their original strength and aesthetic aspect after multiple damaging events. Specimens built with calcium-silicate and clay bricks were subjected to subsequent cracking cycles using a crack-mouth-opening-displacement controlled bond-wrench test. Experimental results showed that self-repair, in terms of strength restoration and aesthetic filling of cracks, occurs even after multiple cracking cycles when the self-healing mortar is used with both types of bricks, optimizing the autogenous healing of cement-based mortars. In this context, the healing effectiveness tended to decrease as the crack width and the number of cycles increased. The effectiveness varied also according to the types of brick and healing environment used, e.g. under humid conditions (RH ~ 95%), 50% vs 80% of the original capacity was regained in fully separated couplets made respectively with clay and calcium-silicate bricks. This outcome provides the ground to delineate the remaining testing campaign.

@en

Historical masonry façades are sensitive to various damaging processes. A recent study, looking at the initiation and progression of cracks in masonry, in the range of 0.1 to 5 mm in width and thus corresponding to light damage [1], has allowed for the calibration of finite-element models that include a material model capable of accurately replicating this damage and which is populated with material properties corresponding to existing structures [6]. The models, which also include a soil-structure interaction boundary designed to account for the effect of the soil during earthquake vibrations [7], have been used to determine the fragility of masonry buildings via the proxy of 2D walls [2]. In the study presented herein, the finite element models are employed to replicate the geometry of (historical) masonry facades to determine their sensitivity to light damage as a consequence of the two damaging processes observed to be most common for this type of façade, namely (differential) settlements and (earthquake) vibrations [3]. The masonry façades were first pre-damaged via settlement distortions which generate just-visible cracks in the order of 0.1 mm to 1 mm in width. Then, an acceleration time history corresponding to two different Dutch earthquake events and two recordings of traffic-induced building vibrations [9] were separately applied at the base of the models. In this manner, the effect of existing damage could be assessed in regards to the aggravation generated by vibrations. The settlement part of the study revealed that long façades were more vulnerable to applied soil distortions, for instance. Then, subsequent vibrations further increased damage for intensities measured with a peak ground velocity (PGV) larger than 2 mm/s while the control set of virgin or uncracked façades remained undamaged at this PGV. At 32 mm/s, many pre-damaged façades also exceeded the light damage range. At equal PGV, the traffic vibrations, with a larger number of effective cycles, resulted in increased damage aggravation in comparison to the earthquake recordings.

@en

This study aims to investigate the damage response of unreinforced masonry (URM) façades resting on strip foundations and subjected to ground settlements via numerical models. The models depict the non-linear constitutive behaviour of both the masonry, via smeared cracking, and of the soil-foundation interaction, via nonlinear interface elements. The influence of building features, such as the masonry material, the length over height (L/H) ratio of the geometry, the wall thickness, the number and size of openings and different types of strip foundations (i.e. reinforced concrete and unreinforced) is examined. A sensitivity study additionally investigates the influence of the interface stiffness and its constitutive model. A Gaussian curve is used to replicate the shape of the ground settlements; These simulate the loss of support underneath the foundation due to urban subsidence. Eight settlement shapes are applied in the FE models, including both symmetric and asymmetric profiles, while the angular distortion is used to measure their intensity. A new aspect is that the extent of the induced damage to the façade is assessed objectively using a damage parameter that represents the number, length and width of cracks in a single scalar value. The method distinguishes between the applied settlement profile at the bottom of the interface and the retrieved settlement profile measured on the façade. The analyses indicate that for a value of the angular distortion equal to 2 ‰ (or 1/500), computed from the resulting deformations of the façades, 60% of the models exhibit serviceability damage associated with cracks of about 5 mm width. Accordingly, the limit values available in the literature are observed to be too optimistic and not conservative in relation to the analyses presented in this study. A key outcome is that facades with an L/H smaller or equal to 1 do not exhibit cracks wider than 1 mm. Façades on reinforced concrete foundations were observed to be less susceptible to settlement damage, compared to unreinforced ones.

@en

The probability of light damage to unprepared, unreinforced masonry structures exposed to induced seismicity due to gas extraction in the north of the Netherlands is still under investigation. Repeated light seismic excitations caused by frequent, light and nearby earthquakes have been linked to economical losses and societal unrest in particular, with extensive damage claims. Moreover, the damaging potential of the seismic events has been related to the condition of the structure, especially if damage corresponding to settlement causes is already present. A comprehensive testing campaign oriented towards the initiation and progression of light damage of replicated clay brick masonry has been conducted at Delft University of Technology. Based on these tests, calibrated finite element models have been produced. This article uses the calibrated non-linear time-history models to simulate the effect of earthquake ground motion on a variety of initial conditions, wall geometry, material properties, and number, type and intensity of earthquakes. The results are then used to regress a relationship between damage and these parameters. This is subsequently employed to run a MonteCarlo simulation and produce fragility curves where the probability of exceeding specific damage values for various initial damage levels is presented against the seismic hazard. The vulnerability or fragility curves show that visible damage, with cracks wider than 0.1 mm, appears, with a 10% exceedance probability, at 13 mm/s of peak ground velocity; but, if the masonry had already undergone some light, yet imperceptible damage, a PGV of 6 mm/s was sufficient to aggravate it into visible cracks. To attain a 1% probability of exceeding light damage however, for which the masonry would need more invasive repair, it was observed that PGVs larger than 15 mm/s were required. These fragility curves were finally compared to graphs from other authors and found to capture well the variability in the range assigned to light damage.

@en

Recent, light earthquakes induced by the extraction of gas in the north of the Netherlands have been linked to light, mostly aesthetic damage of the traditional masonry structures in the region; this is also connected to economic losses and societal unrest. To be able to accurately assess the light damage, detailed finite element models are necessary and need to include realistic soil movement, wave propagation, and soil-structure interaction boundaries. Moreover, the minute deformation of the soil, including the rocking and translational components of seismic ground motion, has shown to be influential to light damage. Consequently, this study has pursued the definition of efficient soil-structure interaction boundaries to implement in finite element models of buildings. A methodology, following the sub-structure method for the seismic Soil-Structure-Interaction (SSI) is defined and presented. The soil-structure-system is divided into three sub-systems: the far-field soil, the near-field soil and the superstructure. First, a 3 km deep and 8 km wide, plane-strain model of the soil is employed to study the behaviour of the soil at the surface due to deep, simplified seismic events. The soil model is linear-elastic since only light seismic excitations are considered. Next, a smaller, 30 × 300 m (shallow) soil model with a building on top, is given boundary elements calibrated to replicate the behaviour observed at the surface in the larger model. Finally, 2D models of masonry façades set on the intermediate soil model are used to reduce the soil-structure interaction to representative interface elements. The models are matched in terms of dynamic behaviour, strains, cracking, and displacements, and the behaviour is compared to existing ground motion data for the Zeerijp and Westerwijtwerd earthquakes. It is demonstrated that the equivalent interface allows efficient modelling of seismic excitations considering a detailed soil-structure interaction for complex, smeared non-linear, time-history analyses of wall models to assess (light) damage in probabilistic studies. Models with this equivalent interface show greater damage than comparison models without it.

@en

Recent, induced earthquakes in the north of the Netherlands have led to a large number of damage claims. Many claims can be considered to fall into the category of ‘light damage’ to the ubiquitous, unreinforced masonry structures in the region. To evaluate and predict the behaviour of cracks, characteristic of light masonry damage, caused by seismic or other actions, an experimental campaign, linked to the validation of computational models, has been pursued. To accurately capture the initiation of visible cracks, wider than 0.1 mm, Digital Image Correlation (DIC) was applied to monitor the entire surface of full-scale wall panels and smaller specimens. Moreover, an optimised speckle pattern and solving algorithm was developed to be able to monitor not only the initiation, but also the propagation of the cracks during subsequent (repeating) loading cycles. In this approach, the crack data is then used to characterise the intensity of damage with a single scalar; the parameter, denoted Ψ and comprising the number, width and length of the cracks, is used to evaluate the progression of light damage in experiments and finite element models. A description of the DIC technique applied and of the development and usage of the damage parameter for masonry is presented herein.

@en

To pragmatically answer the question of whether masonry walls can withstand a hydrostatic water pressure, a clay-brick wall was built and tested at Flood Proof Holland. The specimen was exposed to various hydrostatic loads, also in combination with debris impact loads, with the goal of providing experimental data for the calibration of structural models and later development of fragility models to study and assess (economic) damage and life loss from potential Dutch floods. The 2.7x2.7 m2 wall, 100 mm thick, was constructed on a steel rig and subjected to a water level difference of up to 125 cm, where the level, at what was considered the back of the wall, was kept low (0 cm). The wall, which was restrained on all four boundaries by being glued at the bottom and top steel beams and constrained against the lateral steel columns of the rig, was subjected to various combinations of water level in front and behind it with the most unfavourable combinations leading to minor cracking. The out-of-plane deformation, perpendicular to the face of the wall, reached approximately 2 mm for the hydrostatic pressure produced by 95 cm of water. The masonry structure, which was also subjected to a vertical overburden by the steel rig mimicking the vertical stress produced by gravity loads on buildings, deformed in two-way bending, meaning that the horizontal supports also prevented the wall from moving. Non-linear, finite element models calibrated against the experimental results offered additional insight into the behaviour of the wall. It was observed that the wall behaved linear-elastically up to a water level of about 90 cm. At higher water levels, the deformation of wall increased significantly which is associated, according to the models, to the initiation of cracking in the wall and the redistribution of the bending stresses from the vertical (stiffer yet weaker) direction to the stronger horizontal direction. At 125 cm, the wall deformation reached up to 5 mm. From hereon, damage in the wall would have progressed, further reducing its stiffness up until collapse at approximately 150 cm of water depth. This latter value could not be achieved experimentally but is inferred from the trend of the experimental results and the understanding provided from the non-linear models. The models explored comprised analytical and numerical models with linear-elastic and non-linear material models, in one- or two-way bending. The behaviour of the wall up until the initiation of cracking could be predicted well with the simpler, one-way bending analytical or FE linear-elastic models, but more complex, two-way bending models were required to output the same fidelity at higher water levels, with only the non-linear, two-way bending FE model capable of representing the experimental behaviour at the highest water levels.
These calibrated models were also used to predict that, if the wall had not been constrained on its sides and thus only at the top and bottom, more alike walls in buildings will long end walls, it would have failed much earlier, at a water depth of about 90 cm depending on the vertical overburden applied. These values for water depth consider that no water was present behind the wall. During the test, water did not infiltrate through the wall, with the highest water pressures leading only to wetness at the back of the wall. In reality, the inside of a building may be subjected to a water level because of water entering the building through windows or doors. The lower differences of water level between the front (outside) and back of the masonry are also associated with lower out-of-plane displacements and lower internal bending moments in the wall. However, in terms of hydraulic head, if the head was located higher and closer to the middle of the wall, it was observed to be more detrimental than an identical head where the inside of the building is dry. The masonry wall was also subjected to a combination of hydrostatic pressure and impact of debris. Two types of debris were considered: a floating tree log representing soft debris, and a suspended steel cube, mimicking hard debris. Impacts of the soft debris of up to 1.8 m/s (90 Joules) did not lead to any additional deformations of the wall, while collisions with the steel cube (up to 180 J) in combination with water depth of about 90 cm, did generate accumulated out-of-plane deformations, visible cracking in a diagonal pattern with leakage of water through the cracks, and ultimately local failure of the masonry. The experiments conducted herein prove that traditional, single-wythe masonry walls in regular buildings can safely withstand water depths of up to 90 cm when the inside of the building is dry. This corresponds roughly to a hydraulic head of 90 cm when the level inside is low. At larger water depths damage in the form of cracks is expected and the boundaries of the wall and its overburden become relevant in determining whether the wall has sufficient capacity to avert collapse. For a square wall, restrained on all four boundaries, the maximum water level was determined to be about 150 cm; but, wider walls or walls without lateral supports are much more vulnerable having fewer possibilities to redistribute stresses, especially to the horizontal direction and thus exhibiting more brittle failure after the onset of cracking. Future work should hence focus on testing the difference between one-way and two-way bending in masonry walls, assessing the effect of leakage through walls with openings, and observing the influence of cavity walls where two walls are built in front of each other with a space in-between, which are common in Dutch buildings, and where water outside and inside the building may exert pressure against the empty cavity. The rate at which such a cavity is filled with water would also provide additional insight into the flood resilience of dutch buildings.@en