Acoustic emissions (AE) are high-frequency (20 kHz - 1 MHz) stress waves created as a result of crack formation and fracture propagation behavior of a material/structure due to quick release of temporary energy from limited resources. AE testing involves recording of such waves using piezoelectric sensors located on the surface of the tested structure and subsequent analysis of the recorded signals to gain an understanding of the nature of their sources. Various parameters exist in relation with acoustic emission signals, including peak amplitude, RMS, counts, cumulative counts, rise time and event time (Figure 1). Such parameters provide vital information for the determination of mechanical and fracture properties of marble. However, background noise should be filtered in order to accurately determine the parameters. Acoustic emission test noise comprises of undesired signals such as 1) mechanical noise, 2) hydraulic noise, 3) electrical (electro-magnetic) noise that are sensed by the transformer. Mechanical and hydraulic noise originates from the uniaxial compression device, while electromagnetic interference (EMI) originates from the AE instrumentation because of electrical transmission and radiation. Electromagnetic noise can be prevented by providing better protection of sensors, preamplifier and main amplifier in metal casings with high conductivity, as well as by earthing casings at a common point. Isolating the transformer in the power source may also reduce EMI.

The low-amplitude, high-frequency acoustic emission wave generated in the marble under load are transformed into electric signals by piezoelectric sensors, after which they are subjected to a series of electronic processes. Subsequent to the data processing phase, which involves amplification, filtering, data collection via an 18 bit, 40 MS/s A/D converter module and DSP-based FPGA, the signal is transmitted to a PC and displayed. Electronic structure of an AE measurement device which runs such processes is exhibited in Figure 2.

Ultrasonic test is performed using receiving and transmitting piezoelectric sensors. Ultrasonic waves (P and S) applied to the test subject by the transmitting sensor are received by the receiving sensor, and the time spent by the wave inside the specimen is measured by the ultrasonic test device. The test device needs to be prepared for measurement beforehand. Such preparation includes entering marble density, pulse voltage, wave type and gain value into the device. Subsequently, the sensors should be placed on top of each other and calibrated. Then, silicone grease should be applied as adhesive on both surfaces of the marble block and on the sensors. Following the time information, which is often determined by direct contact, the device performs the calculations by respectively running the theoretical equations provided in Table 1. Compressional wave velocity is measured first, using data for specimen thickness and the determined time. Then, shear wave velocity is calculated by dividing the compressional wave velocity by two. The μ and σ Lame coefficients are calculated using the density data initially entered into the device and the two aforementioned velocity values. Lastly, the calculated Lame coefficients are used to calculate the Poisson’s ratio, Young’s modulus, bulk modulus, coefficient of compressibility and dynamic elastic constants, respectively.

Mineralogy of the tested material, the shape and dimensions of the spaces-grains and their micro-structural properties, such as distribution, determines the progress of the ultrasonic velocity. Said material properties uniquely alter the frequency spectrum and defines the compressional and shear wave velocities of various types of progress. Therefore, wave velocities and amplitude-frequency spectrums are closely related to the physical and mechanical properties of the material (Poisson’s ratio, Young’s modulus, bulk modulus, efficient of compressibility and dynamic elastic constants), which are in turn closely related to its micro-structural properties.

Ultrasonic test is carried out both under load and without load, the latter being carried out particularly in determining discontinuities in a specimen. Further, the surface needs to be polished in order to achieve good ultrasonic transmission. Acoustic emission similarly involves ultrasonic wave detection. However, in that case, the sound source is the material itself and the signal created by such sound in the sensor is analyzed. On the other hand, ultrasonic test involves simultaneous assessment of the receiving and transmitting signals. Figure 3 exhibits the application of the ultrasonic test on cylindrical and prismatic specimens, and Figure 4 demonstrates a comparison of ultrasonic test outputs via AE.

LVDTs are a widely used type of transformers which can transform linear movements of an object, to which it is mechanically connected, into electric signals. These are able to measure small movements of various magnitudes, ranging from several millionths of an inch to several inches. Components of an LVDT able to detect positions up to ±30 inches (±0.762 meters) are shown in Figure 5. Internals of a transformer comprise of primary winding with a pair of identically wound secondary windings which are positioned symmetrically on both sides of the primary winding. The coils are wound on a monobloc hollow form of thermally stable glass-reinforced polymer, which is encapsulated against humidity, covered by a highly conductive magnetic shield and fixed on a cylindrical stainless-steel casing. This coil setup is usually the fixed component of the position sensor.

The moving component of an LVDT is a separate pipe-shaped armature made of magnetically conductive material. Being referred to as the core, this component can freely move axially inside the empty bore of the coil and is physically connected to the object the position of which is to be measured. Typically, the bore is large enough to provide a significant radial distance between the core and the bore, and there is no physical contact between the coils. In operation, the primary winding of the LVDT is energized with alternating current with appropriate amplitude and frequency, which is referred to as primary excitation. The electric output signal of the LVDT is the differential AC voltage between the two secondary windings, which varies based on the axial position of the core inside the LVDT coil. Said AC output voltage is often transformed into a higher-level DC voltage or current, which is more suitable for use in the relevant electronic circuits.

The primary winding of the LVDT, P, is energized by a constant-amplitude AC source. The magnetic flux thereby created is connected by the core to the adjacent secondary windings, S1 and S2. If the core is located halfway between S1 and S2, then it is connected to both secondaries with equal flux, in which case the voltages induced in the S1 and S2 windings are equal to E1 and E2, respectively. Differential voltage output (E1-E2) is essentially zero at this reference middle position of the core, which is therefore referred to as point zero. As shown in Figure 6; if the core is moved to a position which is closer to S1 than S2, then more flux would connect to S1 than to S2, which would cause an increase in the induced voltage E1 and a decrease in E2, which, in turn, would create a differential voltage (E1-E2). In contrast, if the core is moved closer S2, then more flux would connect to S2 than to S1, which would cause a decrease in E1 and an increase in E2, which, in turn, would also create a differential voltage (E2-E1).

An atomic force microscope (AFM) scans the surface of a specimen using a console with a very sharp tip. As the tip approaches the surface, the close-distance attractive force between the surface and the tip causes the console to deviate towards the surface. On the other hand, when the console gets closer to the surface in a manner that would cause the tip to contact it, a progressively increasing repulsive force comes into play, causing the console to deviate away from the surface.

A beam of laser is used to detect the deviations of the console towards or away from the surface. Any deviation of the console would cause small variations in the direction of the beam which is reflected from the flat top of the console. A position-sensitive photodiode (PSPD) can be used to track such variations. Thus, if an AFM tip passes over an elevated surface attribute, then the resulting console deviation (and the subsequent change in the direction of the reflected beam) is recorded by the PSPD.

The console of an AFM scans over the region of interest and displays its topography. Elevated and dented attributes on the specimen surface effects the console deviation which is monitored by the PSPD. Utilizing a feedback loop to control the height of the tip above the surface, thereby preserving a constant laser position, AFMs are capable of creating an accurate topographical map of surface attributes (Figure 7).

Surface scan by an AFM is achieved not by the movement of the console but by the movement, in the x, y, z directions, of the specimen placed on a piezoelectric material. Piezoelectric materials (piezo crystals) are ceramic materials that could enlarge or contract when voltage is applied. This allows the specimen to make very precise movements in the x, y, z directions (the position can be controlled in resolutions in nanometer levels).

Environmental scanning electron microscope (ESEM) is a direct descendant of the conventional SEM, but also allows scanning of wet and non-conducting specimens without any prior specimen preparation. A low gas pressure (up to approximately 10 torr) can be applied around the specimen. When water is used instead of gas, hydrated specimens can be kept in their natural state. Be it water or any other gas, ions created as a result of the collisions of the electrons emitted from the specimen and the gas molecules help reducing the creation of charges and are dragged back to the specimen surface. This eliminates the need to apply conductive surface coating to non-conductive specimens (Figure 8(a)). Such two significant benefits of ESEM makes the strengths of the scanning electron microscope (SEM) available for a wide range of materials. Additionally, a comparison of the operating conditions of ESEM and SEM is given in Table 2.

In conventional scanning electron microscope (SEM) systems, the surface of a solid specimen is excited by a highly focused energetic electron beam, which induces X-ray fluorescent from the elements comprising the specimen. Specimens need to be operated under high vacuum conditions and need to be conductive or coated with a conductive material, such as gold, in order to ensure proper analysis. In conventional SEMs, X-ray micro-analysis is prevented for non-coated non-conductive specimens due to specimen loading which could reduce the energy of the electron beam and renders quantitative analysis impossible. Environmental Scanning Electron Microscope (ESEM) is a unique system which allows analysis of non-coated biological and industrial materials, such as marble, by an electron beam in a water vapor atmosphere at high room pressure. Thus, specimens can be analyzed using ESEM without the need for destruction or any additional preparation procedures. The main drawback of a typical ESEM, however, is that the electron beam propagates in the high-pressure environment chamber and excites fluorescent X-rays not only under the beam but from the entire specimen. Fluorescent X-rays emitted from outside the area of interest are

also sensed by the detector and reduces image contrast. Polycapillary optics can be included into an existing ESEM in order to collect X-rays from an area defined by the focal point dimension and focus the same onto the detector, thereby reducing background noise and improve image contrast (Figure 8(b), Figure 8(c)).

Both other electron microscope types (SEM and TEM) are equipped with energy-dissipating X-ray spectrometers. They are a strong tool to analyze chemical composition of a specimen. Each element emits an X-ray at a certain wavelength, its “fingerprint”, which is detectable and analyzable. This allows matching of the elements on the sample area at an atomic resolution. SEM releases electrons, which scatter upon impact, and the specimen surface is thereby scanned. The device gathers the scattered electrons and creates an image. The image is displayed on a TV-like screen (Figure 9(a), Figure 9(b)). On the other hand, TEM processes the specimen by sending an electron beam. The result is displayed using a fluorescent screen. SEM images are three-dimensional accurate representations, while TEM images are two-dimensional and require some interpretation (Figure 9(c), Figure 9(d)). However, TEM has an advantage over SEM in terms of resolution and zooming (Table 3).

Micro-CT is a 3D imaging technique utilizing X-rays to see inside an object, slice by slice. In a Micro-CT system, X-rays are generated in an X-ray source, transmitted through the specimen, and recorded by the X-ray detector as a 2D projection image. The specimen is then rotated, and another X-ray projection image is taken (Figure 10). This step is repeated through a 180-degree turn (or sometimes 360 degrees, depending on specimen type). The series of X-ray projection images is then computed into cross-sectional images through the computational process called “reconstruction”. These slices can be analyzed, further processed into 3D models, made into movies, printed into 3D physical objects, and more.

DEM (discrete element method) is a numerical technique for modeling the interaction between individual particles and boundaries in order to predict the behavior of cast solids. This device can easily model moving boundaries and used to better understand particle flow dynamics. The DEM method uses contact detection algorithms and can calculate the forces acting on particles by applying relevant contact models. Acceleration, velocity and position are then calculated using Newton’s laws of motion and digital integration. The industry-leading software implementing the discrete element method (DEM) used to simulate the movement of granular and discontinuous material is ANSYS Rocky (Figure 11).

In the Proton Nuclear Magnetic Resonance (Proton NMR) method, a specimen is placed in a strong magnetic field and subsequently subjected to impact by radio frequency radiation (Figure 12). The protons will have their spin states aligned or not aligned with the magnetic field, and the difference between such two states will create a resonance energy. When hit with the right frequency, protons with lower energy will reverse their alignment. As a result, what is detected using this technique is the energy emitted by those protons which revert back to the lower energy state (Figure 13).

Laser Ultrasonics is a contact-free ultrasonic analysis technique based on excitation and ultrasonic measurement using two lasers. A laser pulse is sent to the test specimen, creating an ultrasonic pulse which propagates throughout the material following interaction with the surface. The ultrasonic vibrations are measured using a laser vibrometer and displayed using a digital oscilloscope (Figure 14).

Polarized light microscopy is a contrast-enhancing technique which can be used to analyze the structure of an object (Figure 15). Polarized light microscopes have higher sensitivity and can be used in quantitative and qualitative research studies. Polarized light microscopy is often used to analyze the anisotropy of a specimen’s optical properties caused by stretching and deformation of initially isotropic materials. The specimens may be crystalline materials (pigments, minerals, etc.) and fibers as well as environmental particulate and biological materials. Polarized light microscopes are conventional standard microscopes equipped with two additional polarizing elements (polarizer and analyzer) which allow the specimen to be analyzed under polarized light. Polarized light microscopes allow observation of specimens that are visible primarily due to their optically anisotropic characteristics. Anisotropic materials have optical properties that vary with the crystallographic axis and direction of the incident light. Such materials have multiple refractive indices (e.g., azurite, cinnabar, etc.). However, these are polarization colors (not true specimen colors) and are related to the specimen’s refractive indices and thickness. Isotropic materials (e.g., gases, liquids, resins, unstretched glasses, cubic crystals with only one refractive index) cannot be seen under cross-polarized light.

In this paper, we have reviewed the literature for past studies on “crack propagation in marble under stress—mechanical and fracture properties of marble” and each study is compiled and exhibited in Table 4 based on materials, testing setups, test specimens, test parameters and research techniques.

Table 4 shows us that studies in the subject area actively made use of research techniques such as AFM, ESEM, SEM, DEM, CT, LVDT and AE. General results obtained from studies conducted for different axes, marble sizes and crack types can be summarized as below.

Fracture process for each individual marble specimen under different loading and unloading conditions begins with the formation of a distinct damage zone around the crack tip. Subsequently, beginning from the crack tip, the damage zone gradually turns into macroscopic cracks. The cracks propagate and combine. Lastly, rock pieces begin to fall off, the specimen loses its load-carrying capacity and eventually perishes completely [32] .

Cracks formed under different loading and unloading conditions can be broadly classified as wing cracks, secondary planar cracks and secondary inclined cracks. In the case of uniaxial loading, the specimen contains many shear cracks and few tension cracks. Whereas in the case of biaxial loading, most cracks are shear cracks and occurrence of tensile cracks is rare [34] .

Tensile and shear cracks are two different types of cracks involved in the connection process of gradual fractures, as in fault lines. Shear cracks usually occur earlier than tensile cracks and dominate the crack propagation process. However, in static tests, the cracking process is dominated by tensile cracks [41] . When the specimen reaches 78.2% of the peak tension, the critical tensile crack is in its initial stage and the macro fracture actually occurs after the peak tension [45] .

The entire cracking process of defective marbles under dynamic loading can be classified into four main stages: white patches; first shear cracks; secondary tensile and shear cracks; and a combination of macro cracks until eventual failure [51] . Two types of white patches are observed; namely linear and scattered white patches. Linear white patches, which usually develop in the tensile stress concentration area, look markedly different from the surrounding-colored area. A fine-striped crack, which usually develops along the linear white patch, would later develop as an open crack. In contrast, scattered white patches, which develop in the compressive stress concentration area, exhibits an indistinct boundary with the surrounding area. It suddenly develops as a macro-crack around the peak stress [47] .

In Carrara marble, a tensile fracture is characterized by a linear white patch, while a shear fracture is characterized by a scattered white patch. Both shear white patches and tensile white patches are observed before the occurrence of macro-scale shear and tensile cracks. Shear cracks usually occur earlier than tensile cracks, causing the defective specimen to ultimately break with an X-shaped shear zone. Tensile cracks usually develop with a flaw inclination angle (β) of less than 60 degrees. The development of tensile cracks is often limited by the predominant propagation of shear cracks and cannot fully develop into visible open tensile cracks. On the other hand, tensile patches that develop under semi-static loads clearly appear earlier than shear patches [59] .

The damage zone of a tensile fracture includes microcracks with one or more long, dominant, wide elongation deformations surrounded by several microcracks confined within about 300 - 1000 µm (2 - 6 calcite particles). However, in the damage zone of a shear fracture, long dominant microcracks are not observed before fracture [56] . The peak stress and the corresponding strain of defective specimens are lower than those of intact specimens under dynamic loading, and these two parameters generally increase with defect inclination angle while decreasing with defect length.

In addition, in most marble specimens that are under biaxial compression, tension decreases successively until instability occurs. Stress is reduced several times. Moreover, in some cases, stress reduction begins even before it reaches its peak value. Loading speed or strain rate has been proven to affect cracking processes [27] . Increased dynamic loading or cyclic loading leads to an increase in crack density. Cyclic loading includes the processes of blasting vibration, drilling, rock cutting, earthquake, excavation, freezing-thawing and hydraulic fracturing [20] .

In this paper, we have reviewed the literature for past studies on “the effects of freezing and thawing on the mechanical properties of marble” and each study is compiled and exhibited in Table 5 based on materials, testing setups, test specimens, test parameters and research techniques.

Table 5 shows us that studies in the subject area actively made use of research techniques such as CT, LVDT and AE. General results obtained from studies conducted for different axes, marble sizes and crack types can be summarized as below.

Studies show that marble specimens are greatly affected by freeze-thaw (F-T) treatment. As the number of F-T cycles increases, less tensile cracks occur, and more shear/mixed cracks are seen [64] . CT scans show that a complex crack network may occur for a marble specimen subjected to a low freeze-thaw process, and as the F-T cycle grows, the scale of stress cracks decreases, and shear/mixed cracks increase.

Under freeze-thaw conditions, a 9% volumetric expansion occurs when the water turns into ice, and freezing pressure fuels the development of new microcracks in the marble. Cracking speed increases with the increase in temperature and/or relative humidity [64] .

Previous F-T damage greatly affects the mode of cracking. It is argued that tensile cracking mode plays a dominant role for marble specimens not subjected to the freeze-thaw process, and shear cracking mode is for marble specimens subjected to high F-T process. As the number of F-T cycles increases, less tensile cracks occur and more shear/mixed cracks are seen [63] .

In this paper, we have reviewed the literature for past studies on “the effects of water saturation on mechanical properties and fracture behavior of marble” and each study is compiled and exhibited in Table 6 based on materials, testing setups, test specimens, test parameters and research techniques.

Table 6 shows us that studies in the subject area actively made use of research techniques such as CT and LVDT. General results obtained from studies conducted for different axes, marble sizes and crack types can be summarized as below.

Wet marble can be up to 40% weaker than dried marble. It can be argued that such difference in strength is due to expansion that increases the high pore fluid pressure and due to the help of grain boundary shift (GBS) in contributing to creep [67] .

The effect of water on the strength of calcite marble at high temperature is small. For instance, the strength difference between wet and dry marble at 700˚C is less than that at lower temperatures and does not exceed 10% on average.

The bending of calcite marble panels should be considered as irreversible expansion of the rock under the effect of temperature changes and humidity. Humidity is considered as a key parameter related to the phenomenon of bending and can lead to unlimited residual expansion in certain types of marble. In general, if the panel is exposed to sunlight and high humidity only from the outside, it starts to bend outwards; whereas if there is moisture behind the panel in a confined space where relative humidity is close to 100%, it starts to bend inwards [68] .

It is still unclear how moisture combined with temperature cycles (day-night) causes marble to deteriorate. It is assumed that average wind speed is the primary factor that causes flexural differences.

In this paper, we have reviewed the literature for past studies on “the effects of temperature on the micro-structure and physico-mechanical properties of marble” and each study is compiled and exhibited in Table 7 based on materials, testing setups, test specimens, heat treatment and research techniques.

Table 7 shows us that studies in the subject area actively made use of research techniques such as SEM, LVDT, AE and UT. General results obtained from studies conducted for different axes, marble sizes and heat treatment methods can be summarized as below.

Environmental temperature fluctuations create new microcracks in marble and cause existing ones to expand, resulting in an increase in porosity, which is the first stage of decay. It has been shown that numerous factors contribute to marble erosion due to thermal changes. These include mineralogical composition of marble, its grain size and shape, texture and type of junction between crystals [83] .

The effects of heating on two different marble specimens are not the same. For example, the volume increase in black marble (consisting of crystallized calcite) with heating is due only to new fracture formations, while in white marble (consisting of microcrystalline calcite) heating causes minor deterioration and a larger increase in volume (55%). The effects of heating are determined by the mineralogical and textural properties of the marble. For crystalline marble specimens (black marble), formation of new cracks is much more important than matrix distortion, whereas the opposite is true for micritic limestone (white marble). In addition, thermal expansion causes more damage in calcite marbles than in dolomite marbles [81] .

When a marble specimen is heated to 200˚C, 400˚C and 600˚C and subsequently cooled to room temperature (25˚C), its color changes considerably due to exposure to high temperatures and many microcracks occur in the marble [78] . As the applied temperature increases, longitudinal wave velocity, uniaxial compressive strength and Young’s modulus gradually decrease and the peak stress corresponding to the peak force increases [77] .

It was discovered that P wave velocity decreases with the increase in process temperature, and effective porosity increases with process temperature. It was also found that Poisson’s ratio gradually increases with temperature [72] .

Temperature changes from room temperature to up to 60˚C have no significant effect on tension-strain curves. However, when temperature reaches 800˚C, the stress-strain curve is concave [75] .

The largest expansion and the largest separation between grain boundaries occur at 90˚C temperature. Thermal decomposition at temperatures above 200˚C has been found to cause formation of significant microcracks in marble [84] . It has been found that the number of thermally-induced microcracks grows significantly with the increase in the number of heating and cooling cycles. However, it was determined that the width and length of such microcracks increase substantially with the increase in the number of heating and cooling cycles [71] .

A rise in process temperature gradually increases the effective porosity and total porosity while significantly decreasing the ultrasonic P wave velocity, which indirectly reflects thermal cracking [76] . In addition, the velocity of ultrasonic pulses moving in marble depends on properties such as density, texture, anisotropy, saturation, temperature and elasticity, as well as porosity and pore size [74] . The directional dependence of the total size of the thermal expansion coefficient and permanent deformation is highest in marbles with strong texture, while Carrara marble with weak texture exhibits a uniform crack formation [84] .

In general, strength properties and acoustic and thermal conductivity of marble gradually decrease with the increase in temperature, while permeability and electrical conductivity increase instead [76] . In addition, heating increases ductility to a certain extent.

In this paper, we have reviewed the literature for past studies on “determination of discontinuities in marble sites” and each study is compiled and exhibited in Table 8 based on materials, observation depth, test specimens and research techniques.

Table 8 shows that the GPR research technique has been actively used in studies in this field. General results obtained from the studies conducted in different regions, depths and areas can be summarized as follows.

Uncontrolled discontinuities in marble quarries are the main factor that reduces efficiency. There is no way to change them. However, if the thickness of the marble to be extracted from the potential quarry and frequency distances of the discontinuities (fractures, cracks, cavities) it contains are determined prior to the opening of the marble surface (mirror), this would significantly reduce operating costs and unforeseen losses. In such case, it is estimated that a block efficiency of 10% and more can be obtained from the 1st region, 5% to 10% from the 2nd region less than 5% from the 3rd region. In addition, it was determined by studies that the commercial size block volumes that can be extracted from regions 1 and 2 would vary between 1.20 m³ and 7.90 m³ [86] .

In this paper, we have reviewed the literature for past studies on “crack assessment in marble sculptures” and each study is compiled and exhibited in Table 9 based on materials, testing setups, test specimens, heat treatment and research techniques.

Table 9 shows that the UT research technique has been actively used in studies in this field. The general results obtained from the studies carried out for different axes, sculpture types and heat treatment methods can be summarized as follows.

The so-called “granulation” of marble is a very common phenomenon of deterioration which results from environmental temperature fluctuations and affects historical monuments and modern buildings alike. Thermal cycles are in fact responsible for the formation of microcracks at the boundaries between calcite grains, which expose the marble to granular fragmentation [92] .

Degradation of marble is mainly related to the excessive anisotropic thermal expansion of carbonate crystals [90] . Artificially eroded specimens produced by heating fresh Carrara marble at 400˚C for 1 hour exhibit widespread microcracks at grain boundaries, just like naturally decomposed samples.

The most frequently used technique in these studies is the measurement of the velocity of longitudinal ultrasonic waves via the direct transmission method. What various ultrasound techniques all have in common is the utilization of a single-channel electronic system with transmitting and receiving transformers having frequencies lower than 350 kHz. Measurements are time-consuming and the locations or dimensions of internal defects can only be estimated [91] .

It was determined that microcracking and increased porosity in marble sculptures result in the ultrasonic pulse rate to decrease from 5.1 to 1.9 km/h after heating. A decrease in V_p can be explained by an increase in porosity. Further, the V_p velocity is also dependent on the mineralogical, physical and mechanical properties of marble, the degree of water saturation and the level of degradation. These minor changes in the pore structure of the sculptures are eliminated by consolidating with hydroxyapatite, which provides good durability in terms of resistance to thermal weathering, salt crystallization cycles and F-T cycles. V_p velocity can thereby be increased from 1.9 km/h to 3.6 km/h. The most alarming cracks in sculptures occur in the “broncone,” which is defined as the trunk of the right leg [88] .