DEM modelling of the activation and reactivation of capable faults in a typical Apulian rock succession: the viewpoint of local seismic effect during the 1948 Earthquake (Apulia, Italy)
Article Category: Original Study
Published Online: Feb 14, 2025
Page range: 1 - 16
Received: Oct 10, 2023
Accepted: Dec 16, 2024
DOI: https://doi.org/10.2478/sgem-2025-0001
Keywords
© 2025 Bruno Giovanni et al., published by Sciendo
This work is licensed under the Creative Commons Attribution 4.0 International License.
A wide portion of the Italian territory is at high seismic risk because of both the high number of earthquakes recorded and the magnitude of some of them. This risk is due to several factors such as high population density in some areas, heterogeneity in building types, and presence of historic-monumental buildings. However, the main factor remains the basic seismic hazard of the territory. The majority of earthquakes recorded in Italy are of tectonic nature; however, some earthquakes of volcanic origin are localised in specific areas of the territory and generally of low magnitude. This means that these earthquakes are caused by the formation and/or reactivation of fault ruptures as a result of intra- and inter-plate movements.
Even the Apulian region, with its role as the foreland of the Apennine orogen to the west and the Dinaric-Ellenic orogen to the east, is not exempt from earthquakes. Seismic events are often from seismogenic zones east of the Adriatic Sea and sometimes, as in the case study discussed here, generated in seismogenic areas of active and capable faults that fall within the Apulian foreland.
Fault ruptures generate seismic waves and cause a zone of deformation called the
Constitutive elements of a
The extent and magnitude of brittle and plastic deformations induced by fault rupture propagation (damage zone) have a significant impact on the permeability of the rock mass. This generally results in increased permeability and vulnerability to the transport of pollutants released to the surface, especially in the case of karst underground aquifers [8], [9], [10], [30], [47]. Of utmost importance are the effects of LSR that occur in inactive faults and, above all, during the activation/reactivation of active and capable faults [53], [57], [44] on the integrity of anthropic surface structures [29] and on the coseismic effects that can be generated (landslides, subsidence, displacements, etc.).
In previous studies [44], [45], [46], [49], [50], the lithotypes modelled had homogeneous and isotropic geotechnical behaviour unlike the lithoid and anisotropic lithotypes used in the present study. Furthermore, a generic time history of accelerations was used for the dynamic analyses, which is not compatible with the expected seismicity in the geological–structural context of the case study, as required by the building regulations (NTC18) in Italy [52]. Finally, no LSR assessment was carried out on the ground surface involved in the fault damage zone. The use of appropriate numerical codes and properly parameterised geomechanical models is indispensable in understanding the mechanisms governing such phenomena as well as in defining their magnitude and areal extension. Therefore, in order to overcome the limitations posed by previous studies, it was decided to carry out DEM numerical simulations of the deformation and LSR phenomena associated with the formation or reactivation of a capable fault in a stratigraphic sequence consisting of competent and anisotropic rock masses such as those typical of the Apulian area.
The structural setting of the case study area [60], [61] is characterised by the presence of a fault system, the most famous of which is known in the literature as the ‘Candelaro’ fault, which forms the dividing boundary between two geodynamic domains, the foreland domain to the east, consisting of the Gargano promontory, and the Bradanica foredeep domain to the west, consisting of the ‘Tavoliere delle Puglie’ plain. The system consists of several NW–SE-oriented faults that have undergone polyphase tectonics during the Mesozoic to Quaternary period [24], and currently, they exhibit a kinematism of right trans-tensional fault with a predominantly normal component [58], [63]. Of these faults, the one considered for the case study (pink in Fig. 2) consists of two stretches, displaced by the extensive W–E transcurrent, right-handed shear system of the ‘San Marco in Lamis-Mattinata’ fault.
Schematic geostructural map of the ‘Candelaro’ fault area (from: [27] modified).
Lithostratigraphically, the foreland domain in this area consists of an approximately 350-m thick bedrock of carbonate platform rocks referable to the Jurassic-Cretaceous [27], [60]. It consists of light grey and light brown limestones and dolomites in 0.30- to 2-m thick layers plunging southeastward at a 10–20° dip. In transgression on the bedrock are Plio-Pleistocene calcarenites, which are about 40-m thick and are known in the literature as ‘Gravina limestones’ [60]. These are yellow-ochre biocalcarenites with variable degree of cementation and massive appearance. In the foredeep domain of the Apulian Tavoliere Plain, above the Gravina Calcarenites there are sub-Apennine Clays of the Pliocene sup.- Pleistocene inf. which are followed by marine and continental regressive deposits of the Pleistocene that constitute different orders of terraces.
The stratigraphic succession consisting of a bedrock of faulted Mesozoic limestones covered by Gravina calcarenites is typical not only of the case study area but of the whole Apulian foreland. For this reason, it was chosen as representative to implement parametric numerical modelling and simulation of the local seismic response in the case study of reactivation of the ‘Candelaro’ fault.
Previous studies on the propagation of fault plane rupture have been conducted with numerical simulations [4], [44], [46], [49], [51], [62]; physical model tests [11], [20], [43], [44], [50]; and
The case study fault area is affected by intense seismic activity both historically and instrumentally (Fig. 3). In particular, the seismic sequence of August 18–22, 1948, with quakes on August 18th (21:12 GMT), 21st (08:45 GMT) and 22nd (23:16 GMT) and epicentre located on the ‘Candelaro’ fault, recorded a macroseismic intensity (MCS) Imax= 6.5–7 and moment magnitude Mw= 5.3–5.5 [33], [34]. This fault in accordance with commonly accepted definitions in literature studies [36], [37] is to be classified as active and capable. As a matter of fact, it was definitely active until the upper Pleistocene [40] and it is capable of being reactivated by the right shear system of the composite seismogenic sources ITCS058 – San Marco in Lamis-Mattinata and ITCS003 – Ripabottoni–San Severo, which divide it into two trunks [28]. Evidence that the ‘Candelaro’ fault is capable of creating ruptures on the topographic surface is the seismic sequence of the year 1948 that caused extensive damage and opened a 12-cm wide, 50-cm long and about 3-m deep fissure in the ground in ‘contrada Ripa’ of the town of Rignano Garganico from which gas escaped [55].
Observational data from field studies and theoretical studies suggest that an earthquake magnitude (Mw) can correlate with the seismic moment (Mo) and the amount of displacement (D) along the causative fault by means of the following equation [35]
The aforementioned equation proposed by Hanks and Kanamori [35] was implemented using values of the shear modulus Gi = 25GPa, already used for the case study rock in previous numerical modelling [14], [16], [18], [35] and usually assumed for crustal faults in the literature [64], [65], and a moment magnitude Mw = 5.5 that characterises the seismicity of the ‘Candelaro’ fault case study [33], [34]. Consequently, it was possible to estimate that in the case of a remobilisation of the ‘Candelaro’ fault by an earthquake of magnitude Mw = 5.5, the value of the average displacement expected along the fault plane would be D = 0.92 m. This value is a useful reference for evaluating the accuracy of the results of the numerical analysis of LSR.
The accelerograms used for the dynamic and LSR of the case study analysis were obtained with REXEL software [38] from the European Strong-motion Database [2], [3]. Consultation of the database was done using the search parameters given in Table 1 and respecting the seismo-compatibility criteria according to the current Italian seismic building code (NTC18) [52].
Input parameters for searching accelerograms in the seismological European Strong-motion database.
| Horizontal | 5.6–7 | 0–20 | A | T1 | 50 | II | SLV | No |
| Vertical | 5.2–7 | 0–30 | A | T1 | 50 | II | SLV | Yes |
The average spectra of horizontal and vertical accelerations of seismic motion, used as an input for the dynamic and LSR analyses, were plotted together with the elastic spectra for rigid substrate of cat. ‘A’ and horizontal topography according to the current Italian seismic regulations (Fig. 4).
Average input spectra of horizontal and vertical accelerations used for LSR analysis, compared with elastic spectra for rigid substrate of cat. ‘A’ and horizontal topography [52].
Numerical simulations were conducted on a geomechanical model consisting of a rigid layered bedrock and a weak, massive, calcarenitic cover (Fig. 5). This stratigraphic succession is typical of the Apulian foreland where the attitude and thickness of the bedrock limestone layers (s = 0.80–1.0 m) were inferred from previous
Geomechanical model used for numerical simulations.
The simulations performed concerned the rupture propagation mechanisms of normal and inverse faults through the bedrock to the ground surface in quasi-static parametric analyses. Moreover, a dynamic analysis with local seismic response LSR evaluation was carried out for the case study of the ‘Candelaro’ normal fault.
According to previous studies [20], [44], [45], [46], [53], [57] in the quasi-static parametric analyses, velocity vectors oriented to generate shear rupture along normal and reverse fault planes, dip at 30°, 45° and 60°, were applied to the edges of the left portion of the undisturbed geomechanical models (black arrows in Fig. 6 a). This first phase of the simulations was preparatory to provide the model used for the dynamic LSR analysis, which focused on the propagation of the fault rupture, under the assumption of seismic reactivation of the normal fault plane in the considered case study. Moreover, the parametric analysis aimed to delineate the trend and magnitude of brittle and plastic deformations along a fault plane that generates and propagates in a succession of lithoid rocks with different elasto-plastic behaviour, as the kinematism (normal and reverse faults) and the dip angles of the faults change.
Models and boundary conditions used in the numerical analyses: a) quasi-static; b) dynamic and LSR (model without pre-existing fault); c) dynamic and LRS (case study with reactivation of a pre-existing normal fault plane 45° dip).
The boundary conditions of the model in Fig. 6a consist of a velocity magnitude of 0.01 m/s (to minimise inertial effects) applied to the several quasi-static analyses. All quasi-static simulations were conducted for a sufficient number of computational steps to create a fault plane with displacements of about 1–3 m in the models. This is consistent with the maximum displacement values estimated to be generated in the case study considered during a single seismic event in literature studies [35], [64] and with the extent of fractures observed in situ after the 1948 summer seismic sequence in the case study of the ‘Candelaro’ fault [34], [55].
Physical–mechanical parameterisation of the lithotypes in the geomechanical model was done using literature data and previously used numerical simulations with the UDEC code [16], [18], [19]. According to the extensive literature on the Plio-Pleistocene calcarenites outcropping in Apulia, the physical–mechanical characteristics (Table 2) are highly variable depending on limestone cement content, grain size assortment and degree of saturation [5], [15], [17], [23], [26], [59].
Minimum, maximum and residual values of calcarenite physical–mechanical parameters (from: [5], [15], [17], [23], [26], [59]).
| Gravina Calcarenite | 19 | 35 – 50 | 15 – 40 | 30 – 38 | 0.13 – 0.29 | 2.0 | 0.11 – 0.26 | 3.5 | 2.9 | 1.3 | 3 – 5 |
In particular, the uniaxial compressive strength value of these calcarenites allows them to be classified as a soft rock [31]. Moreover, due to their strain-softening behaviour, the friction angle, cohesion, tensile strength and dilation angle have been reduced linearly from its peak values to their residual values during a development of 3% plastic strain [19].
Reference was made to data published in literature studies [14], [16], [18] also regarding the physical–mechanical parameters of the Mesozoic limestones, their layer discontinuities and faults (Table 3).
Values of physical–mechanical parameters of limestone, layer discontinuities and faults (from: [14], [16], [18]).
| Altamura Limestone | 24 | 50 | 13 | 69 | 11 | 60 | 33 | 25 | 7 |
| Mechanical parameters of discontinuities for Mohr–Coulomb ‘area-contact’ criterion | |||||||||
| Rock mass | Joint type | Joint normal stiffness JKN (GPa/m) | Joint shear stiffness JKS (GPa/m) | Joint tensile strength Jtens (MPa) | Joint friction angle Jfric (degrees) | Joint cohesion Jcoh (MPa) | Layers dip (degrees) | Layers spacing (m) | Joint dilation angle (degrees) |
| Altamura Limestone | Layers and faults | 249 | 104 | 0.70 | 39 | 5.29 | 0 | 0.80 | 7 |
The analyses were conducted considering deformable blocks, using the code UDEC v.7.00_78 [41], adopting the Mohr–Coulomb strength criterion and strain-softening constitutive models for the calcarenitic lithotype and linear elastic-plastic models for the layered limestone lithotype, respectively. For the fault planes and layer discontinuities of the limestone rock mass, the Mohr–Coulomb criterion and the ‘area contact’ type were adopted. Simulations were performed in terms of effective stress and dry conditions by means of sections with a width–height ratio of 5:1, which is enough to minimise the effects of lateral boundaries on the strain development in the model. To avoid numerical distortion of the propagating wave in the dynamic-LSR analysis, the maximum length of the grid elements should be assumed smaller than 1/10 of the wavelength λmin associated with the highest frequency of the used seismic input motion. For this purpose, all the modelled sections have been meshed via a 5-m maximum length, triangular finite difference grid elements. In the quasi-static analyses (Fig. 6a), a constant velocity oriented as the dip angle of the simulated fault was applied to the mesh boundary of the left portion of the undeformed geomechanical model. At the right portion of the model, on the other hand, zero velocities were imposed in the X-and Y-directions.
The number of calculation steps applied was the same for all parametric simulations. In addition, the velocity increments were in the order of 0.01 m/s, to minimise inertial effects and have quasi-static analyses.
For the simulation of the coseismic ruptures and LSR due to the specific presence and reactivation of the case study fault, two distinct models obtained from the static analyses were used where displacements and velocities were set to zero and only the plasticisation states of the rocks were maintained. The first model (Fig. 6b) has no fault plane; the second, more realistic model has a pre-existing normal fault plane with a 45° dip (Fig. 6c). Supplementary free-field grids coupled to viscous boundaries have been used, to prevent reflection of outward propagating waves back into the numerical model. These consisted in independent dashpots in the normal and shear direction [41]. A 59.99 second seismic waves input was applied at the bottom of limestone bedrock with an incidence angle θ = 0°, and the time histories of velocity associated to the corresponding acceleration records were converted into shear and normal stress time histories (Fig. 7), as usual when using quiet boundaries. Furthermore, a viscous Rayleigh damping with ξmin = 3% and a frequency fmin = 16 Hz were used.
Time histories of the seismic waves input: horizontal ground motion and shear stress (left); vertical ground motion and normal stress (right).
The quasi-static parametric analyses have been conducted on a typical Apulian stratigraphic succession (Fig. 8) by varying the kinematism (normal and reverse faults) and the dip angles of the faults (Table 4).
Displacements, fault core and damage zone extents and plastic states in quasi-static analyses of normal fault generation (left) and reverse fault (right): a–b (dip angle 30°); c–d (dip angle 45°); e–f (dip angle 60°).
Quasi-static analyses: failure types, maximum displacement values, extents of fault core and damage zone.
| Normal fault | 30 | Conjugate failure surfaces | 1.49 | 6.50 | 9.90 |
| 45 | Single failure surface | 1.83 | 7.70 | 9.60 | |
| 60 | Conjugate failure surfaces | 2.58 | 7.80 | 9.50 | |
| Reverse fault | 30 | Trailing imbricate fan and conjugate surfaces | 1.48 | 26.50 | 84.50 |
| 45 | Trailing imbricate fan and conjugate surfaces | 1.82 | 56.10 | 61.50 | |
| 60 | Conjugate failure surfaces | 2.57 | 20.70 | 163.20 |
The results obtained, in terms of type of failure, displacements and fault core and damage zone extents on the ground level, are in fair agreement with those of previous studies [44], [45], [46], [53]. However, it must be pointed out that more of these previous studies were not conducted with a distinct element code but with finite difference codes (FLAC) or finite element codes (ABAQUS) and using models that do not include the layered and faulted substrate in addition to the cover soil (Fig. 6). The distinct-element UDEC code, on the other hand, is well suited for modelling the growth of discontinuities or cover soil deformations under large strains [1], [20], [62]. In addition, this technique can model deformations involving unlimited relative motions of individual elements like strata and rock blocks. In the fault core zone, the propagation of displacements, from the bedrock to the ground level through the calcarenitic cover, deviates slightly from the straight projection of the bedrock fault plane. This phenomenon is most evident in cases of reverse faults, where the damage zone in the calcarenitic cover tends to be more extensive and has a lower dip than the fault plane in the substrate. In all simulations, regardless of the kinematics (normal or reverse) and the dip angles of the faults, the maximum displacements were obtained in the fault hanging wall of the calcarenitic cover (Fig. 8). As for the brittle and plastic failure mechanisms of the rock mass of the limestone substrate and calcarenitic cover, it is observed that they are caused by shear and, subordinately, tensile stresses. In the case of normal faults, the main shear plane tends to split into two conjugate planes as it transfers from the calcarenitic cover to the substrate. This phenomenon is more evident for fault planes that dip at 60° (Fig. 8f, left).
In cases of reverse faults that dip at 30° and 45°, imbricated fan shear structures are generated from the separation boundary between the limestone substrate and the calcarenitic cover and up to the surface of the models (Figs. 8b, d, right). Several conjugate fracture shear planes are also generated, especially in the case of a reverse fault dip at 60° (Fig. 8f, right), which affect the fault footwall of the calcarenitic cover.
The case study concerns the right trans-tensional fault with a predominantly normal component named ‘Candelaro’. The aim is to help shed light on the mechanisms of coseismic rupture, in the case of reactivation of this active and capable normal fault and on the LSR in the projection area of the fault plane onto the ground level. The results of these kinds of simulations have important implications for seismic microzonation of the territory, for the design of new infrastructures and for the safety of existing ones.
The case study dynamic analysis was performed on the geomechanical model as in Fig. 6c, using an input earthquake (dotted blue and yellow lines in Fig. 4), obtained with the software REXEL [38] from the European Strong-motion Database [2], [3]. This earthquake reflects the seismological characteristics of the case study area and is seismocompatible with the 5% damping elastic spectrum of the Italian seismic normative [52].
The maximum displacement obtained, on the order of 0.80–1.30 m (Fig. 9a), occurs deep in the limestone rock mass along the fault hanging wall. Also, it is observed that the failures of the rock mass limestone substrate (Fig. 9b) are caused almost exclusively by shear stresses.
Dynamic case study analysis of the ‘Candelaro’ active and capable normal fault: a) displacements; b) plastic states.
A previous study has shown that soil deformations are not symmetrical [65]. The shear fractures in the limestone substrate are concentrated both along the pre-existing fault plane and along a series of roughly parallel planes to the right and left of it. Interestingly, the main fault plane tends to divide into many conjugate planes with a Graben or flower structure as it moves from the limestone substrate to the calcarenitic cover. Another evidence of the different stress-strain state of the cover rock mass is that its upper portion, on the hanging wall side, is uniformly subject to tensile stresses, while on the footwall side is almost exclusively subject to shear stresses.
Table 5 summarises the values achieved in different ways for the displacements along the fault plane and on the ground level.
Dynamic case study analysis of the ‘Candelaro’ active and capable normal fault: failure type, displacement values, extents of fault core and damage zone.
| Normal fault | 45 | Conjugate surfaces ‘Graben o flower structure’ | 0.80–1.30 | 0.1–0.2 | 0.12 | 0.92 |
Numerical modelling has provided displacements at depths along the fault plane that are in good agreement with those obtained both with the Hanks and Kanamori equation (1) and with geostructural measurements performed in situ on similar normal faults in the Murge area of Apulia [42].
Local seismic response (LSR) analysis, in the presence of layer and fault discontinuities, was conducted with the methodologies commonly used in literature studies [18], [32], [45], [54], [56]. In particular, in order to highlight the contribution on the LSR of the reactivation of a pre-existing fault, simulations were conducted in both the absence (Fig. 10 left) and presence (Fig. 10 right) of a fault plane in the model.
Local seismic response (LSR) of the case study: on the left model without a pre-existing fault plane – a) FA in X-acceleration, b) FA in Y-acceleration, c) plastic states; on the right model of the ‘Candelaro’ active and capable normal fault plane – a) FA in X-acceleration, b) FA in Y-acceleration, c) plastic states.
The amplification/deamplification factor (FA) of the X and Y acceleration components of the ground-level motion was analysed to estimate the effects of LSR. More specifically, in each control point on the ground level of the model (Fig. 6 b, c), the FA has been calculated as a ratio of the values of the X and Y acceleration spectrum in the points at the top of the model and in the corresponding ones at the base of the calcarenitic rock mass for three ranges of vibration periods 0.1–0.5, 0.4–0.8 and 0.7–1.1s. These ranges of vibration periods, normally used in the LSR studies, are also comparable with the natural ones of the most common buildings existing in the case study area.
The LSR trend comparison of the two analysed models allows us to make some considerations about the amplification/deamplification effects at the ground level in an area without a pre-exiting fault (Fig. 10 left) or affected by the reactivation of a capable fault (Fig. 10 right). As can be seen, the LSR of the two models is very different depending on the absence or presence of a pre-existing fault.
In the model without a pre-existing fault, there is a substantial deamplification at the ground surface for all three analysed frequency ranges and for both X and Y motion components (Fig. 10 a, b left). Such deamplification, as already pointed out in previous studies [6], [7], [18], [54], would be explained by the strong dissipation of energy during the plastic deformation of interlayers in the limestone substrate. The model with the pre-existing fault, on the other hand, presents seismic amplification phenomena, for both components of X and Y motion (Fig. 10 a, b right), in the range of vibration periods 0.1–0.5 s and deamplification for the other periods considered. In detail, FA presents a substantially unimodal distribution for X-accelerations and bimodal for Y-accelerations with peak values of FA-Xacc = 2.91, FA-Yacc = 2.22 and FA-Yacc = 1.36, respectively. There seems to be a close correlation between the location of the peak amplification of seismic motion at the surface, in the low period range, the geometry and type of plasticisation zones in the calcarenitic cover with the pre-existing fault. As a matter of fact, the hanging wall zone, which extends about 100 m to the left of the fault plane, has large areas of tensile and shear plasticisation, and it is subject to seismic amplification having the highest FA peaks of both X and Y acceleration components. This particular occurrence could perhaps be explained by the formation of a graben structure, bounded by the fault plane (right) and the plasticisation and shear rupture planes (left). Such discontinuities would result in a kind of constructive entrapment in the reflected/refracted surface seismic waves.
Finally, further confirmation of the accuracy of the physical–mechanical parameters and constitutive models chosen for the dynamic analyses arises from the comparison of the result obtained, in terms of the maximum expected displacement at the ground level, with the numerical simulation (Table 5) and the one calculated using the relationship (2) proposed by the Italian seismic regulations [52]:
Hence, considering FA = 2.91 obtained from the LSR for the X-accelerations of the ground motion, equation (2) gives a maximum displacement value on the ground level equal to dg = 0.12 m which is in good agreement with the displacement on the ground level obtained by numerical UDEC simulation (Fig. 9 a and Table 5).
Numerical modelling and simulations, conducted on a stratigraphic succession typical of the Apulian territory, have provided results that allow some interesting conclusions to be drawn in agreement with the results of previous studies. This series of parametric analyses is applicable to general cases, concerning the propagation of the first fault under dry conditions, from a stratified limestone substrate to the overlying massive calcarenitic cover. The case study simulation, on the other hand, concerns the coseismic rupture mechanisms in the case of seismic reactivation of an active and capable normal fault and the local seismic response (LSR) around the fault plane at the ground level.
The following are main conclusions obtained: parametric analyses with pseudo-static loading
The fault rupture propagation from the bedrock up to the ground level through the cover rock may deviate significantly from the straight projection of the bedrock fault plane. This phenomenon is most evident in cases of reverse faults, where the damage zone in the calcarenitic cover tends to be more extensive and has a lower dip than the fault plane in the substrate. In the cases of reverse faults that dip at 30° and 45°, trailing imbricate fan type shear structures are generated, where the maximum displacement occurs in the shear plane separating the hanging wall from the foot wall (Fig. 8 b, d, on right). Subordinately, there are also conjugate rupture plans. In the case of a reverse fault that dips at 60° (Fig. 8 f), several conjugate rupture planes are generated in addition to the main rupture plane, affecting the fault foot wall of the calcarenitic rock mass. Regarding the failure mechanisms of the limestone substrate rock mass and calcarenitic cover, it is observed that they are mainly caused by shear and, subordinately, tensile stress. The zone of high brittle fracturing (damage zone) and the probably connected increased permeability tend to be concentrated on the hanging wall, in cases of direct fault models, and in the foot wall, in cases of reverse faults. Moreover, its extent is about 6–16 times smaller in normal faults than in reverse faults. The main fault plane tends to divide into many conjugate planes, with Graben or flower structure as it moves from the limestone substrate to the calcarenitic cover (Fig. 9 b). The stress state of the most superficial portion of the calcarenitic cover is not symmetrical. In fact, on the hanging wall side, it is uniformly subject to tensile stress; on the footwall side, it is subject to shear stress. Comparing the results of normal fault dips at 45° for simulations of the quasi-static analyses (Fig. 8 d left) and those of the dynamic analyses (Fig. 9 b), it can be seen that in the former case (conditions of the first rupture of the calcarenitic cover, by fault propagation from the limestone substrate), the fracturing is concentrated almost exclusively along a single plane, which is that of the upward extension of the fault plane. The LSR of the model without a pre-existing fault shows a substantial deamplification at the ground surface for all three analysed frequency ranges and for both Xacc and Yacc motion components (Fig. 10 a, b left). Such deamplification would be explained by the strong dissipation of energy during interlayer plastic deformation in the limestone substrate. The LSR of the model with a pre-existing fault (case study of ‘Candelaro’ fault) shows peaks of ground motion amplification, for both FA-Xacc and FA-Yacc components, exclusively in the range of 0.1- to 0.5-s vibration periods (Fig. 10 a, b right). FA presents a substantially unimodal distribution for X-accelerations and bimodal for Y-accelerations with peak values of FA-Xacc = 2.91, FA-Yacc = 2.22 and FA-Yacc = 1.36, respectively In the LSR of the case study (model with a pre-existing normal fault), there appears to be a close correlation between the position of the peak amplification and the geometry and type of plasticisation zones in the calcarenitic cover. This could be explained by the presence of a Graben structure, bounded by the fault plane (right) and the planes of plasticisation and shear rupture (left), within which a kind of constructive trapping of reflected/refracted surface seismic waves would occur. Confirmation of the accuracy of the physical– mechanical parameters and constitutive models chosen for the dynamic numerical analyses of the case study comes from the good agreement among the displacements predicted at the ground level with the numerical simulation, those measured in situ as a result of coseismic ruptures in the 1948 earthquake and those calculated using equations (1) and (2) widely used in the literature and in the Italian building code (NTC18).
seismic reactivation of an active and capable normal fault and local seismic response (LSR)