In order to understand the cracks movements in historical masonry buildings and their causes, the most important parameter to monitor is the evolution of the width of cracks compared to the evolution of temperature and humidity because they give an idea of the loss of stiffness. Next, some low-cost techniques that could be applied to measure the evolution of cracks width are discussed.
Methods for analogue data collection, such as comparators and callipers, are common techniques to measure cracks evolution. Their main disadvantage is that to reach the most decisive injuries in arch and vault keys, it is necessary to use auxiliary lifting platforms for both installation and reading and control, hence they are quite invasive. These measurement techniques do not leave evidence of the validity of the measurement and therefore lack traceability. In case of doubt regarding a data, due to inconsistency with the series of measurements, it must be discarded. Additionally, the calliper is considered less precise than the comparator, because the device, which must be fixed to both sides of a crack, may give erroneous readings as it can be moved from its initial position by external agents (birds, rodents, …). For these reasons, we explore, next, two low-cost digital techniques to measure cracks evolution: 3D photogrammetry and 2D image-based crack monitoring.
3D photogrammetry is another possible technique to measure cracks evolution. Close-range digital photogrammetry includes a large family of methods based on several acquisitions of a set of images used to produce a 3D point cloud of the scene [14]. Capturing data from the outside using photographic and photogrammetric management is currently having great success in 3D modelling of buildings and archaeological sites, monitoring broken specimens in the laboratory [15], and monitoring of cracks in outdoor infrastructures [16, 17]. Its attempts for indoor measurements (e.g., [12]) present a good quality/cost ratio but require particular attention during the shooting of images in survey campaign [18].
2D image-based crack monitoring is a technique often used to measure cracks in aging civil engineering structures outdoor, e.g. [9, 16]. This technique is considered a solid alternative to other approaches for cracks width monitoring since it ensures the objectivity of measurements, possibility to achieve sub-mm accuracy and the convenience for quickly recorded and stored characteristics of the entire crack pattern, and possibility of permanent observations and off-line measurements performable at any time [9]. As far as the authors of this paper know, this method has never been applied to indoor crack monitoring of historical buildings. The starting hypothesis to select this technique is that to measure the shrinkage or increase of a crack in indoor conditions at a low cost, it is not necessary to deploy a large amount of media, but just to make a 2D measure using a single photo. Barazzetti & Scaioni [19] argue that usually, cracking expands in one plane, so that only one image is enough to reconstruct the 2D geometry. Only in the case of off-plane deformations or non-flat objects, the problem becomes 3D and it requires a stereo-pair of images in order to be solved [19]. In our case, the first movements of the buildings get manifested through cracks in the arches and vaults. These cracks occur in the plane, in 2D, and are the ones that we control in preventive monitoring. Other authors have also proposed the use of a single photo for crack monitoring to reduce costs [16]. They do it by means of indirect measurement systems, using reflective targets for outdoor cracks of infrastructures as reference measurements. However, in this paper, we propose a method for indoor cracks that does not require the use of targets because they would imply the use of lifting platforms, becoming an invasive method for buildings in use. Instead of targets, we look for surface accidents in the vicinity of the cracks, such as the width of the voussoirs, bricks or joints, which we use as references. Barazzetti & Scaioni [19] argue that the measurement of crack deformations between different epochs, as we do in this paper, require the use of permanent targets in order to have permanent references. In our proposal of 2D image-based crack monitoring technique applied indoor to masonry buildings, surface accidents act as permanent references.
In this paper, we first tested the low-cost analogue and digital techniques in a laboratory setting. We tested their accuracy comparing the results obtained with them, using the comparator as a reference. The reason for choosing the comparator as reference is that it is the most broadly used method because it is cheap and accurate. The in-lab study allowed us to identify the 2D image-based method as the most promising technique because it is accurate, low-cost, and does not present the disadvantages of the other techniques considered when applied to indoor cracks of historical buildings with low social projection. Afterwards we checked, in a real situation, its advantages and disadvantages in order to develop standardized protocols of inspection that allows to compare data between different buildings and injuries. As case study we used a church named Nuestra Señora de la Asunción in Cariñena (Zaragoza, Spain), which is currently undergoing a reparation process. In the in-lab test we made use of targets as measurement reference to validate the 2D image-based method, whereas in the in-situ test we used the surface accidents of the construction as reference to avoid the use of targets that would require a lifting platform. Additionally, we have conducted a cost comparison of the different techniques, applied to the case study, in order to show the economic advantage of the proposed technique.
In-lab research materials and methods
In the laboratory of the Mechanical Engineering Department of the School of Engineering and Architecture of the University of Zaragoza, which has controlled temperature and humidity, measurements were made by means of a comparator, a calliper, 3D close-range photogrammetry and 2D image-based crack monitoring, on an artificial crack in a mechanically cut and separated board, following the next indications:
-
Analogue comparator: an IP64 Mitutoyo with 20 mm range and 0.01 mm precision was used, attached to both sides of the crack. Its anchoring system minimizes the possibility of movements unrelated to the model itself or errors for measuring in a different spot (Fig. 1). For this reason, it is considered the most accurate measurement technique, and is used as a reference.
-
Calliper with digital reading: an IP54 Preciva with a range of 150 mm and an accuracy of 0.01 mm was used (Fig. 1). This device introduces a small uncertainty factor when repeating readings, since the width of a crack varies at each point in its length and it is highly unlikely to repeat the reading in exactly the same spot.
-
3D close-range photogrammetry: the camera used was a Canon 200d with 18-55 mm lens, 100 ISO sensitivity and locked Autofocus. A tripod was used to take the photos in order to increase the exposure time depending on the variability of the light conditions. The obtained photos were rendered three-dimensionally using the Agisoft application "PhotoScan Professional". This software tool allows to measure the distance between points of the model created by means of a specific module. The distance to the object was fixed at 12 m. This software has the advantage of including a series of targets (Fig. 1) that, located on the object to be measured, are detected with precision. In this way, once the targets are correctly located, the measurement can be carried out without having a great specialization in metrology. The problem with this technique, as in the case of the comparator, lies in the need to reach the cracks to place the targets. We made a first test of this technique in-lab and also in the Basilica del Pilar in Zaragoza in order to check the quality of the model in indoor conditions. Once done, we discarded the use of the 3D close-range photogrammetry because of its long processing times to create point clouds and meshing and the difficulty of achieving an acceptable model due to the lack of light in indoor spaces, besides the invasiveness that the placement of targets implies. The lack of indoor light could be overcome with reflectors. However, the reflectors installation for 3D photogrammetry is expensive and requires a permit to be placed in a building in use.
-
2D image-based method: the same camera was used at a fixed distance to the object of 12 m. The targets available in the Agisoft application "PhotoScan Professional" software were employed to define the references. With the aim to establish a proper 2D image-based crack monitoring procedure for masonry buildings, we considered the parameters of 3D photogrammetry [20,21,22,23] included in Appendix A to adapt them to our methodology. Then, the crack width was measured. In the literature, to measure the crack width some authors use an algorithm (e.g. [19]) and others use an image processing software (e.g. [16]). Since we had already observed that the image processing software tool for indoor conditions did not provide enough precision, we proposed the use of a drawing application to measure the crack width in indoor monitoring, and we studied the precision it provides. We preferred the use of this method to measure cracks width due to the ease of access and affordability of the tools it requires (camera and drawing application). To choose the most suitable drawing application for photo measurement, three different applications, known and accessible to anyone, were used on a same crack photo: AutoCAD, AutoCAD LT and DraftSight. Finally, AutoCAD LT was selected due to its user-friendliness, speed of loading and precision. The precision of AUTOCAD and AUTOCAD-LT is the same. However, due to its smaller number of modules, LT greatly accelerates its loading on the computer and the creation of new drawings for each crack. This software is available in most architecture and engineering firms. For a cheaper option, DraftSight should be considered. The method consists of comparing the width of the crack with a known fixed distance used as reference at different times to monitor its evolution over time. First, the photo is taken and inserted into a digital CAD file using the AUTO-CAD LT software tool. Then the image is scaled according to the measurement of the known reference distance. Finally, the width is obtained by means of the use of the linear dimension command of AUTO-CAD LT. The sequence of steps to measure cracks width with AUTOCAD LT is included in Appendix B. Since the support material used for the artificial crack is a wooden board, the reference distance practically does not undergo expansion. This 2D technique provides several advantages which make it an interesting method for crack evolution monitoring, such as no need for lifting platform and auxiliary materials to reach the furthest points, non-invasiveness with the use of the building, standard equipment for any technician, possibility of repeating the measurement in the same position and spot, and traceability. The latter is important, since it allows to reviewing the accuracy of the data obtained at any moment, and this is useful for future historical monitoring. However, this technique remains unexplored for this specific purpose, probably due to the increasing popularization of automatic monitoring.
Once the techniques to be compared were decided (comparator, calliper and 2D image-based crack monitoring), they were tested in two measurement ranges to evaluate their precision, in mm and hundredth of mm.
To compare the three techniques in the mm range, we used a cracked wooden chipboard (Fig. 1), which was subjected to a linear mechanical separation by steps of approximately 2 mm: from step 4 mm to step 20 mm, with separations at 2 mm interval. The first and last steps were eliminated for providing the most aberrant data. The two boards slide on a rail, allowing to maintain the parallelism of the crack. In the upper part targets were arranged to be able to measure the crack evolution using the 2D image-based technique. In the middle part, an analogue comparator was attached to both sides of the crack. In the lower part, a calliper was installed to measure the opening of the crack. Once the position was fixed, a picture of the whole was taken. To reduce the uncertainty of the calliper measurement, marks were drawn on the chipboard to perform each measurement at the same crack spot.
The usual cracks have a different width along their length and an angular shape. In our test, thanks to the equal separation throughout the entire crack and lack of the angular shape, we got a better comparison of the different measurement methods. We used the measurement offered by the analogue comparator as reference, and the calliper for validation as well as to be able to mechanically increase the crack by steps of approximately 2 mm. Measurements were taken in one day without humidity or temperature variation. In order to store the data, a model inspection sheet was prepared, and an example of its use can be seen in Fig. 2.
Afterwards, the comparator and the 2D image-based technique were compared in the tenth of a mm range. The crack was immobilized by fixing two steel plates above and below the measurement points at the back side. In this case the measurement stand is a wooden support, so there can be practically no expansions or retractions due to temperature, as compared to the high dilatation coefficient of the steel plates. Subsequently, the movement of the crack was measured due to the expansion of the steel plates with the variation of temperature, assuming negligible that of the wood. After a measurement of the crack in the initial position of 23.68 mm and calibrating the comparator in 0 mm, on 19th December 2018, the period without heating due to the Christmas holidays was used to allow contraction and subsequent expansion of the plates once the heating service resumed. To measure temperature and humidity, a commercial thermohygrometer (Oh! haus & co brand) was used. In order to store the data, another model inspection sheet was prepared, and an example of its use can be found in Fig. 3. As can be observed, the 2D image-based technique shows the distance between the targets used as reference, whereas the comparator shows the crack width increase.
In-situ research materials and methods
The building used as case study was the Church of Santa María de la Asunción in Cariñena, facilitated by the archbishopric of Zaragoza. It presents injuries in the keystones of the arches on which the dome rests.
The first visit took place on June 14th, 2018. Permission was requested to introduce a lifting platform to place analogue comparators in the cracks against which to test the 2D image-based measurements. As it is a building in use, this permission was never obtained. This is how we realized that the methodology for crack monitoring should not involve the use of targets, since obtaining permissions to install lifting platforms in this type of buildings in use is a cumbersome and time-consuming task. In the Church of Santa María de la Asunción in Cariñena, we only applied the 2D image-based crack monitoring technique, after we had selected it and evaluated it in lab, in order to verify its efficiency in a real building.
The dome of the Church had suffered a lightning strike, on the 29th of June of 2016 which caused injuries to the lantern, dome and support arches. They had been repaired and hidden under the plaster and paint in March 2017. In addition, two tie rods were arranged in the arches 2 and 4 perpendicular to the naves (Fig. 4), which are the ones with the least horizontal offset in the supports. Subsequently, the visible cracks were grouted with elastic filler and the whole was painted. Therefore, our monitoring serves as validation control of the repair.
Four cracks were monitored in the keystones of four arches. To take the photos, the camera was placed under the arch opposite to each of the cracks, at a distance of 25 m between the camera and the crack. The positions of the camera can be seen in Fig. 4, where the photo taking points are indicated with a camera logo. Additionally, the direction towards which photos were made is indicated graphically and by means of an alphanumeric code (P#) whose number corresponds to the code given to each photographed keystone crack (Arch #). For example, P1 is the photo taking point, from which the keystone crack of Arch 1 was photographed. A pavement mark was used to place the camera in the same place all the visits. This distance of 25 m offers an angle close to 45º to better see the arch change in plane vertical-horizontal. Small variations in successive shots should not be considered because the proportion between the crack measurement and the reference measurement will always be the same regardless of the angle and distance to the object. Once the photo is scaled by means of the reference, we obtain the measurement of the crack. To ensure that the width variation is measured at the same point, the change of plane between the vertical side and the intrados of the arch is chosen, due to its ease of observation and differentiation as can be seen (Fig. 5).
The camera used again was the Canon 200d with 75-300 mm zoom lens, 100 ISO sensitivity and locked Autofocus. A tripod 1.5 m high was used to take the photos to allow increase automatically the exposure time depending on the variability of the light.
The fact of having arches and vaults with painted mortar lining increases the difficulty of determining the measurements and location of the cracks, as they are masked by a continuous layer on the brick joints. If they were visible, the joints of the bricks could be used as reference distance, with which the precision of the result in the photographic measurement models is higher.
The ideal would be to have a reference of reduced size to facilitate the measurement of the crack and to minimize the expansion movement of the reference distance. In our case the reference distance was of 0.89268 m (Fig. 5). It is the distance to the closest crack. This high distance implies a higher uncertainty because it can suffer higher dilations due to temperature. To assess the uncertainty due to the high reference distance, we considered the dilation of the reference distance in relation to a dilation of the arch. If the arch has a width of 10m, the length of the perimeter is 39.27 m. If due to the retraction of the arch, the crack has a decrease of 3 tenths of a millimetre, i.e. 0.3 mm, it means that the dilation of the arch has a coefficient of 0.000007. For the reference distance considered, 0.8926 m, it would involve an increase in the size of 0.00000625 m, i.e. 0.000625 mm. This is a low uncertainty since it is outside the range of 0.1 mm, considering a crack is severe when its width is over 2 mm [24, 25]. For this reason, we validated the reference distance.
It is considered that if a measurement is carried out annually and the building does not have problems, in the same month the next year the width of the crack and the size of the reference distance will be substantially the same for the same temperature of the arch. If the building has no heating system, as is the case, this temperature depends on the average outside temperature of the previous month due to the thermal inertia of the arch. For this reason, it is important to determine the average outside temperature of the previous month. This allowed us to verify at the end of the monitoring process, that it is substantially the same than the previous year.
The pendentives had moisture spots due to lack of waterproofing, appearing to be the origin of the deformation of the arches. To control the humidity of the pendentives, a thermographic control was carried out that would make it possible to locate variations in the accumulation of water by comparison with an accessible area of a wall moistened by capillarity. Thermography allows to detect a temperature scale from which the existence of thermal bridges caused by moisture becomes visible. A termography camera (Flir brand) was used to take temperature photos of the arches. To measure the moisture of the base wall, a moisture meter (Meterk brand) was used with a range of 2% to 70%. These measurements were made systematically in dry and rainy periods monthly throughout the crack measurement period.
To manage the cracks data collection of each arch, a model sheet was created (Fig. 5). The collected data included the average exterior temperature of the previous month, the interior temperature at the time of the measurement, the interior relative humidity, and the surface humidity on a wall due to capillarity. In this way, it could be discerned if the movement of each crack was due to the variation in temperature or to the cession of the supports because of humidity.