ANALYSIS OF THE IMPACT PROCESS AT DIKES WITH CROWN WALLS AND PARAPETS

This paper is focused on the analysis of the impact process at dikes with crown walls and parapets under breaking and non-breaking waves. A small-scale laboratory campaign was performed at the Hydraulic Laboratory of Bologna. The experiments were aimed to analyze the vertical pressure distribution along the crown wall and the resulting wave forces, by varying geometrical and hydraulic parameters. The tested configurations included different off-shore slopes, dike crest widths, crown-wall heights, dike crest freeboards and the inclusion of the parapet. The measurements were combined with the image analysis of the run-up and of the wave impact process. A sub-set of the experiments was numerically reproduced, with the openFOAM modelling suite, to support and to extend the experimental results. The results confirmed the link between the air content, the shape and the magnitude of the pressures according to the breaker type, already observed for larger-scale experiments.


INTRODUCTION
A promising solution to upgrade coastal and harbour defense structures, to face the effects of climate change, is represented by the use of crown walls (Burcharth et al., 2014), eventually with parapets (Van Doorslaer et al., 2015;Formentin and Zanuttigh, 2019). While a wide literature is available on the combined effects of the wave breaking and of the air entrapment on the wave impacts at seawalls (Plumerault et al., 2012), few works are dedicated to dike-type structures with crown walls.  and Castellino et al. (2018) performed small-scale experiments and numerical modelling to evaluate the effect on the vertical pressure distributions of parapets placed on top of vertical walls under non-breaking waves. The results showed that the inclusion of the parapet increases the pressures along the crown wall due to the impulsive pressures enhanced by the confined return flow. Van Dooslaer et al. (2015, 2017) and De Finis et al. (2020 confirmed the same result in case of parapets on the crown walls of smooth dikes. Their research was primarily focused on the reduction rate of the overtopping discharge at the rear side of the structure, under non-breaking waves only. Later, Zanuttigh and Formentin (2018) extended this database by testing similar structures also under breaking waves. The gathered data are here used to calibrate a numerical model, developed in the openFOAM environment, aimed to support and extend the laboratory investigations. Indeed, the reliability of numerical models in representing the wave impact dynamics, and the so the wave forces acting against the crown walls, was deeply investigated (Ma et al., 2014;Liu et al., 2019). However, these analyses were often focused on single and/or regular wave impacts, characterized by small air content.
Therefore, this paper aims at analyzing the effects of the air entrainment and of the model scale on the magnitude and on the distribution of the wave pressures at crown walls and parapets of smooth dikes, under both breaking and non-breaking waves. Specific objectives are i) to verify whether the findings by Bullock et al. (2007) for large-scale experiments are still valid at small model scale, and ii) to validate a reliable numerical model to support the physical investigation.
The paper is organized as follows. Section 2 presents the methodologies adopted, which implied both the physical and the numerical investigations. Section 3 describes the main results obtained during the laboratory campaign. Section 4 show the numerical model investigations, describing the calibration and the analysis of the wave impact, giving a qualitative indication about the air content associated to the tested breaker types.

METHODOLOGY
This work is based on the results of a new laboratory campaign on wave overtopping recently conducted at the University of Bologna. The new tests were aimed at collecting an extensive and systematic experience on the wave impacts at crown walls with and without parapets. The analysis of the pressures acting against the structure is supported by image analysis to derive the link between the air content, the magnitude of the loads and the shape of the pressure signals. The numerical model, developed in the openFOAM environment, was calibrated based on the experimental work, to correctly reproduce the hydraulic and the structural performance of such structure, and to derive qualitative indications about the air content at the wave front.

Laboratory campaign
The laboratory campaign was performed at the Hydraulic Laboratory of the University of Bologna in the wave flume, that is 12 m long, 0.5 m wide and 1.0 m deep. It is provided by a piston wave-maker, capable of generating both regular and irregular wave attacks. The water depth h at the wave-maker should not exceed 0.4-0.45 m. The wave flume is provided with a recirculation system, composed by i) a recirculation conduit, ii) a pump and iii) a flowmeter, responsible of maintaining the water depth constant during the experiments. The characteristics of the wave conditions that can be implemented are: a maximum significant wave height Hs of approximately 0.06 m and a maximum wave length Lm-1,0 of ≈3 m. All the laboratory tests involved irregular wave series, characterized by a Jonswap spectrum with a peak enhancement factor γ=3.3. The tested wave conditions included wave heights Hs in the range of 0.05-0.06 m and spectral wave periods Tm-1,0 in the range of 0.85-1.45 s, giving two reference values of the wave steepness sm-1,0≈0.03 or 0.04.
Several instruments were installed inside the wave flume and across the dike structure to measure the time series of the free-surface elevations and the pressures. Specifically, 3 resistive wave gauges (wgs), were placed at approximately 1.5•times the maximum Lm-1,0, to obtain information about the incident and reflected waves, according to Zelt & Skjelbreia (1992). A fourth wg was installed on the crest width to measure the thickness of the overtopping layer. All the wgs are characterized by a sample frequency of 100 Hz. To quantify the wave impacts, 3 pressure transducers were installed along the crown wall, characterized by: sample frequency of 1 kHz; range of measurement from 70 mbar to 700 mbar; accuracy of ±0.04% full scale; external diameter of 25 mm, internal diameter of 3 mm for the measurement of the pressures. The run-up and the wave impact processes were filmed by means of a full HD camera (30 fps) positioned outside the wave flume, in correspondence of the crest width. The full scheme of the wave flume, with reference to the position of the wave gauges, can be found in Zanuttigh & Formentin (2018).
The tested configurations consisted of smooth dikes with a crown wall at the inshore crest. The wall might include or not an inclined parapet. The layout of the typical cross-section, with reference to the main hydraulic and structural parameters, is given in Figure 1. Specifically, α is the constant dike offshore slope, Gc the crest width, hw the height of the crown wall,  the inclination of the parapet, and Ac and Rc the distance between the dike crest and the end of the crown wall from the still water level, respectively. The experiments were carried out in 1:20 scale and consisted of 128 irregular tests. The smooth slopes (cotα=2 and 4) are characterized by different crest levels (Ac/Hs from 0 to 0.5) and crest widths (Gc=0.15 and 0.30 m). The combination of wall heights (hw=0.04 and 0.05 m) and water levels (h) are such to determine relative freeboards Rc/Hs in the range 0.67-1.50. The angle of the parapet, when present, was fixed ε=30°, based on previous analyses (Zanuttigh & Formentin, 2018). The combination among the wave and the structure characteristics provide an Iribarren-Battjes breaker parameter ξm-1,0 in the range of 1.23 and 4.0, including in the database both breaking and non-breaking waves.

Numerical investigation
The 2D numerical simulations were performed by means of a 2DV RANS-VOF software, i.e. openFOAM (OF). Specifically, the toolbox waves2Foam, originally developed at the Technical University of Denmark by Niels Gjøl Jacobsen et al. (2012), was used for the wave generation/absorption inside the numerical channel. It is a solver capable of solving 2 incompressible, isothermal immiscible fluids using the Volume Of Fluid (VOF) phase-fraction based interface capturing method. This library is a modification of the native solver interFoam, based on the Volume Average Reynolds Average Navier Stokes equations (VARANS). The fluids air and water are simultaneously tracked using the scalar field , which is equal to 0 for air and 1 for water. Intermediate values represent a mixture of the 2 fluids. In the momentum balance equation, an extra term is included to take into account of the surface tension between the 2 phases. The wave generation/absorption processes occurred by means of the application of the relaxation zone technique (active sponge layers), providing a large range of wave theories for both regular and irregular wave series.
The numerical simulations reproduced a sub-set of the laboratory wave conditions at the same laboratory scale. The irregular wave series, characterized by a Jonswap spectrum with a peak enhancement factor γ=3.3, are implemented by defining the values of Hs and Tp. The sub-set was selected to evaluate the variation of the forces acting on simple crown walls, under breaking and non-breaking waves, by varying the significant wave height Hs, the wave steepness sm-1,0, and by keeping constant the crest width Gc and the height of the wall hw. Table 2 summarizes the numerically tested parameters. The length of the domain was set equal to 11.2 m, which corresponds to more than 3 times the wavelength L characterizing the tests (≈3 m). It was provided with an inlet relaxation zone for the wave generation, equal to 4.5 m, followed by 2 wavelengths to let the wave to be completely developed before reaching the structure. The position of the offshore edge of the crest with, and so of the crown wall, was kept constant for all the tests (10.7512 m and 11.0512 m, respectively-x coordinates). The dike footprint depends on the cotα values, and so on the length of the sloping part, i.e.1.72 m for cotα=4 and 1.02 m for cotα=2.
The numerical domain dimensions were set to correctly represent all the wave conditions. The mesh characteristics slightly change along the domain to minimize the computational effort, while maximizing the accuracy of the results. Specifically, the numerical domain was divided into 3 parts along the horizontal direction, with a cell size dimension of 0.05 m-0.002 m, 0.002 m and 0.002-0.02 m, respectively. The second zone is the one related to the run-up/wave impact processes. Therefore, a very refined constant grid was preferred. In the vertical direction, starting from the bottom, the cell size varies from 0.01 m to 0.002, in correspondence of the still water level, to reach the value of 0.05 m in correspondence of the atmosphere. The patches composing the numerical domain, which have to be characterized by a boundary condition for each initialized wave field, i.e. alpha.water (VOF), p_rgh (dynamic pressure), U (velocity), are summarized in Table 3. The numerical domain, shown in Figure 2, was equipped with 3 gauges, located in the same position of the laboratory wave flume, to measure the incident and the reflected waves (see Figure 2). In the numerical model, the number of the pressure transducers was increased by setting them every 0.005 m from the base of the crown wall, i.e. 9 numerical pressure transducers (see Figure 3). The average overtopping discharge rates q were computed by integrating the horizontal water velocity components (output every 0.005 m) along a virtual gauge placed at the rear side of the crown wall. Figure 3 shows a detailed scheme related to the pressures and the velocity-VOF outputs aim to quantify the forces F and the overtopping discharge rates q, respectively.

LABORATORY ANALYSIS AND RESULTS
This Section presents the analysis performed on the laboratory campaign results. The results on the air entrainment derived from image visualization/analysis and on the reconstruction of the vertical profiles of the wave pressures from the transducers are in agreement with the literature available for large-scale experiments (Oumeraci et al., 1993;Bullock et al. 2007). Indeed, a strong correlation is found among the breaker types, the amount of air pockets entrapped and the type and the entity of the impact loads.

Image analysis and pressure signals
The literature presents different classifications methods of the wave impacts and the wave pressure signals, according to different aspects to be considered. Table 4 reports the classification of the tested wave conditions and the resulting pressure signals according to 3 of the most common methods: 1) the method based on the Iribarren-Battjes breaker parameter ξm-1,0, which classifies the pressure signals according to the nature of the wave breaking (surging, plunging, broken); 2) the method proposed by Oumeraci et al. (1993), aiming at characterizing the nature of wave impact (impact load, slightly breaking); 3) the method by Bullock et al. (2007), based on the air content associated to the wave impact (high aeration, low aeration). Table 4. Classification and description of the tested wave conditions according to the breaker parameter, the type of the wave impact and the associated air content.
Breaker parameter (Iribarren-Battjes, 1973) Wave impact (Oumeraci et al. 1993) Air content (Bullock et al, 2007) Surging (non-breaking, the wave front reaches the wall fully developed) Impact load High-aeration (broken, the wave front breaks against the wall) Plunging (slightly-breaking, it represents a transition phase between the non-breaking and the broken wave

Slightly breaking
Low-aeration (slightly-breaking) In this laboratory campaign, the wave impacts were classified according to the breaker type. Based on the values of ξm-1,0 calculated at the toe of the dike slope, the tested wave conditions involved: surging (ξm-1,0>2.0), plunging and broken (ξm-1,0<2.0) waves. The main difference is that the former breaks during the impact with the crown wall, while the latest two reach the structure during a transition phase (plunging/slightly breaking) or when they are already broken. Figure 4 shows a few frames of different wave impact types. In panel a), the non-breaking wave type reaches the crown wall before breaking; in panel b), the wave is breaking just at the end of the run-up process, overturning and hitting violently against the crest; the flow is characterized by the first air bubbles beneath the wave front. In panel c), the wave is already broken when completing the run-up process and when propagating along the crest width. The flow shows a high turbulence rates and huge air entrainment amount. In panel d), relative to the crest freeboard Ac=0, the wave is completely broken before reaching the top of the slope, and the flow along the crest is bore and low-energetic, due to the dissipation occurred during the wave run-up, presenting a significant level of air bubbles, which can be hardly individually detected. For all the tested conditions with Ac=0, the breaker types are always identified as broken; in case of Ac>0 instead, the breaker types can be either plunging (≈80-86%) or surging (≈7-10%) according to the value of ξm-1,0 associated to each incident wave. The Iribarren-Battjes classification, combined with the freeboard crest conditions, identified thus 3 wave types in the laboratory campaign. As an example, Figure 5 shows the signals associated to a surging (a), a plunging (b) and a broken wave (c) occurring consecutively during the same test (Hs=0.05 m, sm-1,0=0.03 Ac/Hs=0,Gc=0.30 m,cotα=2,hw=0.5,ε=30°) at the pressure transducers P1 (in blue), P2 (in orange) and P3 (in yellow). By looking at these pressure signals obtained from the present small-scale laboratory campaign, it is possible to observe the same findings of the large-scale experiments performed by Oumeraci et al. (1993) and Bullock et al. (2007). Indeed, the surging wave pressure signals (a) show all the characteristics of a high-aeration impact load: the ratio between the maximum peak and the quasihydrostatic peak, pmax/ph,q, is greater than 2.5 for all the pressure transducers and the signals are characterized by strong oscillations due to the huge air content inside the wave front, with a subatmospheric peak after the first expansion phase. The second impact (b) shows the shape characteristic of the slightly breaking impact (plunging wave): in this case, the ratio pmax/ph,q varies between 1 and 2.5 as for the large-scale experiments (Oumeraci et al., 1993;Bullock et al., 2007) and the pressure signal is smoother and the frequency of the oscillations is lower due to the smaller air content. The last impact (c) is classified as broken, with a ratio of pmax/ph,q ≈1 and oscillations denoting a huge presence of air, which in this case is in form of bubbles due to the earlier breaking process. This latter impact presents on average a frequency of occurrence of ~17%, which becomes close to ≈100% in case of Ac>0, when the wave breaking always occurs along the dike slope during the wave run-up phase. Though all the tests with Ac=0 present both plunging and surging breaker types, the most frequent impacts correspond to plunging waves (≈85%), especially in case of cot=4 (>90%), when ξm-1,0=1.23-1.94. The surging breaker type is more frequent (≈13%) in case of cot=2, with values of ξm-1,0=2.38-4.03.

Geometric and hydraulic parameter effects on the pressure profiles
This section presents the pressure vertical profiles, for both breaking and non-breaking waves. Note that the adjectives "breaking" and "non-breaking" are here used to refer to the target wave conditions of the tests based on the target values of ξm-1,0 respectively < and >2, and not to the characteristics of the single wave impacts. Indeed, as already specified and as shown in Figure 5, each test presents both breaking and non-breaking waves. Into specific, the analysis shows the effects of the geometric parameters, combined with the wave conditions, on the vertical pressure distributions. The reference value is the statistical pressure p250, which is the average of the highest N/250 impact events, where N is the number of waves of the test time series. It is widely used in the literature, being more representative of the wave impact dynamic with respect to the maximum pressure. Therefore, for each tested configuration the vertical profiles were reconstructed by computing the dimensionless values of p250/(ρgHs) in correspondence of the pressure transducers installed along the crown wall, where ρ is the water density, g the gravitational acceleration and Hs the significant wave height.

Figure 6. Example of the vertical pressure distributions in case breaking wave condition, with reference to the effect of the variation of Ac/Hs (panel b), sm-1.0 (panel c) and Hs (panel d). All the tests refer to structures with cotα=4 and hw=0.05 m.
By increasing the crest width Gc (light green lines in Figure 6), there is a strong reduction of the wave pressure along the crown (≈ 60-70%, on average), which is more pronounced if combined with the parapet (dashed lines in Figure 6). However, the inclusion of the parapet does not induce a systematic effect on the pressure trend, as it is highlighted by the average non-dimensional values of p250/(ρgHs) in case of crown walls (1.53, 1.80 and 1.20 at P1, P2, P3, respectively) and wall with parapet (1.60, 1.78, 1.20). The variation of the crest freeboard Ac/Hs strongly affects the entity of the wave loads. Indeed, the emerged cases (Ac/Hs = 0.5) show a reduction of the values of p250/(ρgHs) in the range of 15-100%, with respect to the crest at the still water level (Ac/Hs=0). Furthermore, in emerged conditions (Ac/Hs=0.5, see panel b in Figure 6), the shape of the vertical distribution presents a triangular distribution, recalling a hydrostatic-shape distribution. This is due to the fact that the number of wave impacts decreases from P1 to P3, lowering the values of p250/(ρgHs) towards the upper part of the crown wall (both with and without the parapet). The wave steepness sm-1,0 and the wave height Hs seems to play pure scale effects. Indeed, higher values of sm-1,0 and of Hs induces a reduction and an increase of the wave loads reduction ( Figure  6 panel a vs. panel c), respectively. Figure 7 shows representative vertical profiles of p250/(ρgHs) for the non-breaking configuration, with hw=0.04 m and cotα=2, presenting more frequently surging waves (≈13%) than other configurations. In each panel of Figure 7, the values of Hs and sm-1,0 and Ac/Hs are kept constant, showing the effects of the variation of Gc (0.15 and 0.30 m in light and dark orange shading, respectively) and the presence of the parapet (dashed line instead of continuous). The 2 panels highlight the effect of increasing Ac/Hs from 0 (panel a) to 0.5 (panel b). The values of sm-1,0 and Hs (which are respectively equal to 0.03 and 0.05 m in both the panels) are not considered because there is no relevant difference with respect to the case of plunging waves (see Figure 6), while Gc does not play a systematic role in the reduction of the wave loads, differently from the previous cases. However, the main difference between the plunging and surging wave conditions is represented by the effect of parapet that strongly increases in the latter case the wave loads acting along the wall (50-70% on average, reaching and exceeding in some cases of the 100%  (Kortenhaus et al., 2003;Castellino et al., 2018), where the difference between the peak pressures in the 2 cases is up to 2 times for breaking waves and even 10 times for non-breaking waves.

NUMERICAL MODELLING RESULTS
This Section presents the results obtained by means of the numerical investigation performed on a sub-set of the laboratory experiments. Firstly, the model was validated based on the average overtopping discharge rates q and on the reflection coefficients Kr, to assess its capability of reproducing the laboratory tests. Secondly, the numerical model was adopted to analyze the forces and the air content associated to the wave impacts. Table 5 reports the characteristics of the tests reproduced by means of the numerical model. Specifically, it reports the ID that identifies the tested case, the freeboard of the crest width Ac, the freeboard of the crown wall Rc, the values of cot, the significant wave height Hs, the peak and the spectral period Tp and Tm-1.0, the wave length Lm-1.0 and the relative freeboard of the crown wall Rc/Hs. The calibration subset was selected to account for the variation of the wave steepness sm-1,0, the significant wave height Hs and the freeboard condition of the crest. Figure 8 and Figure 9 show the comparison among the laboratory (subscript lab) and the numerical (subscript mod) results in terms of the average overtopping discharge rates q and of the reflection coefficients Kr, respectively. The values of qmod show a good agreement with the laboratory results (Figure 8), despite of the different methodologies used to quantify the average overtopping discharge rates. Indeed, the values of qlab are reconstructed from the overtopping volumes, while in the numerical model the horizontal velocity components, combined with the VOF values, are integrated along the vertical gauge, placed at the rear side of the crown wall (see Figure 3). The values of Kr are slightly underestimated by the numerical model (see Figure 9). Eventually, to quantify the reliability of the numerical model, 3 different errors were computed, i.e. the relative error (Eq. 1), the RMSE (Eq. 2) and the Wilmott index (Eq. 3). The performance is high when the first 2 parameters go to 0, while the third goes to 1. The results are shown in Table 6, for both the values of q and Kr.

Numerical analysis of the wave impacts
The numerical model was adopted to investigate the wave impact dynamics. As for the pressures, the hydrodynamic forces F were treated as stochastic values, and the statistical values of F250 were computed as the average force value of the highest N/250 impact events, where N is the number of waves of the tested time series. Figure 10 shows the comparison between the experimental and the numerical values of F250/ρgRc 2 . The numerical model slightly overestimates the forces measured during the laboratory campaign, for both breaking and the non-breaking waves. Specifically, the higher are the forces, the higher is the discrepancy. Although several studies are available for the analysis of wave impacts against rubble mound crown walls (a.o. Franco et al., 2018;Jacobsen et al., 2018), the literature related to crown walls placed on top of smooth dikes is very limited. Van Doorslaer et al. (2017) was the first to propose a design formula to predict the wave forces, under irregular wave series, based on non-breaking waves only. Such formula considers the non-dimensional F250/ρgRc 2 , predicting an exponential decreasing trend of the wave forces with the relative freeboard Rc/Hm0 (Eq. 4), where b is equal to -2.02 and -2.4 for small-scale (1:10, 1:15) and large-scale (1:6) tests, respectively. Figure 11 shows the comparison of the numerical values of F250/ρgRc 2 , with the formula by Van Dooslaer et al. (2017), presented in Eq. 4. The pressure signal analysis highlighted that the most frequent impact in the present investigation (80-86%) is the slightly breaking impact, i.e. a plunging wave that represents a transition phase between the surging (pure nonbreaking) and the broken wave (pure breaking). The numerical forces show a good agreement with the formula developed by Van Dooslaer et al. (2017), despite of the different model scale. The numerical model was used to quantify the air content along the crown wall, in correspondence of the virtual pressure transducers (Figure 3). The required outputs are related to the Volume of Fluid (VOF) values, which varies from 0 (air) to 1 (water). Table 7 reports the percentage of the air content measured during the numerical simulations VOFair,%. As expected, the highest percentages were registered towards the top of the crown wall, which is rarely reached by the wave front. Indeed, lower values are measured only for the case of Hs=0.06 m that represents the highest simulated wave height. The most interesting aspect can be observed by comparing the zero-freeboard case R00H05s3G30c4W4 with the corresponding one characterized by an emerged freeboard crest, i.e. R05H05s3G30c4W4. In fact, even if the still water level is lower than the freeboard crest, the incoming waves always hit the crown wall, even if with lower intensity. Therefore, the highest values of VOFair,%, registered in the lower part of the crown wall, are due to the presence of air bubbles, caused by the early breaking process. The same results can be observed for the tests characterized by a cotα=2 (Table 7).

CONCLUSIONS
A new small-scale laboratory campaign, involving 128 tests, was performed at the Hydraulic Laboratory of Bologna. The experimental analysis aimed to analyze the wave overtopping and the wave impacts at dikes with crown walls and parapets. Different structure configurations were tested under irregular wave attacks, including both breaking (ξm-1,0≈1.23-2) and non-breaking waves (ξm-1,0≈1.23-2). The tested configurations included different dike slopes, crest widths and freeboards, crown wall heights with or without a top parapet.
The specific objective of the investigation was to perform a parametric analysis of the effects of the structure geometrical parameters on the wave impacts acting on the crown walls. The preliminary analysis of the image frames and of the pressure signals allowed to associate the tested configurations and the breaker types. As for large-scale experiments (Oumeraci, et al., 1993;Bullock et al., 2007;Plumerault et al., 2012), the magnitude of the impact pressure and the shape of its signal is strongly dependent on the breaker type and the amount of air pockets entrapped. Indeed, the most violent impacts observed are associated to the non-breaking wave conditions, characterized by the presence of small air pocket. In case of breaking/broken waves, the crest width Gc significantly contributes to reduce the magnitude of p250, up to 60-70%. Therefore, the increase of the crest width might represent an effective solution to reduce the enhanced loads due to the parapet inclusion on the top the crown wall. In case of non-breaking waves, the introduction of the parapet induces a severe increase of the values p250, i.e. 50-70% on average; while the effect of Gc is negligible. Therefore, the inclusion of parapet in case of structures subjected to surging waves is not recommended.
A numerical investigation was performed on a sub-set of the experimental tests. The model calibration was performed based on the overtopping discharge rates q and the reflection coefficients Kr obtained from the experimental campaign. The accuracy of the model was assessed by means of 3 error indices, i.e. the relative error, the RMSE and the Wilmott index. For the values of q the relative error, the RMSE and the Wilmott index are -9.21%, 0.0009 and 0.95, while for Kr, -16.40%, 0.09 and 0,90, respectively. The statistical values of F250 are slightly overestimated by the numerical model. However, considering that the pressure signal analysis highlighted the slightly breaking impact as the most frequent in the present investigation (80-86%), the numerical results were compared with the formula developed by Van Dooslaer et al. (2017), tuned on similar structures under non-breaking wave conditions. The analysis performed shows a good agreement between the theoretical formulation and the numerical results, despite the different model scales. The numerical model was used to qualitatively assess the air content VOFair% in correspondence of the virtual pressure transducers. As expected, the top of the crown wall registered the highest values of VOFair% because less frequently reached by the waves. Indeed, these values resulted to be lower in case of Hs=0.06 m, the highest tested significant wave height. The tests characterized by an emerged freeboard crest shows high values of VOFair% at the base of the crown wall, with respect to the correspondent cases with the crests at the still water level. This result indicates a huge presence of air bubbles in the wave front, due to the earlier breaking process that occurs along the offshore slope. Further research will focus on the validation of these data with the support of image analysis to quantify the air content according to the breaker type and its direct consequence on the magnitude of the pressures and so the forces.