Comparative Analysis of Asphalt Core and Clay Core Earthfill Dam Under Varied Earthquake Loading Conditions

Abstract

Earthfill dams located in seismic regions are highly vulnerable to earthquake-induced deformations, particularly when founded on soft alluvial soils. This study presents a comparative numerical investigation of earthfill dams with asphalt and clay cores subjected to seismic loading. A 20 m-high zoned embankment dam founded on soft alluvial deposits was modeled in PLAXIS2D and subjected to four earthquake records. The dynamic responses at the crest and downstream slope were evaluated in terms of acceleration, settlement, and lateral displacement. The results indicate that while lateral displacements are nearly identical for both core types, dams with clay cores experience significantly higher seismic settlements, reaching up to 35% more than those with asphalt cores under strong earthquake loading. Overall, the asphalt core demonstrated enhanced resilience, exhibiting reduced settlement due to its higher stiffness, viscoelastic behavior, and inherent capacity for self-healing following seismic loading.

1. Introduction

One of the oldest known embankment dams is the Sadd El-Kafara Dam in Egypt, constructed around 2600 BC [1]. Since then, earthfill dams have been widely used worldwide for water storage, irrigation, hydropower generation, and flood control due to their adaptability to diverse geological conditions and cost-effectiveness.

Many existing earthfill dams are located in seismic regions, where their performance is strongly influenced by both material properties and foundation conditions [2]. Even under static conditions, inadequate design may lead to excessive settlement and pore water pressure development [3,4]. Under seismic loading, additional hazards arise, including excess pore pressure generation, stiffness degradation, and loss of stability, which may result in significant deformation or failure [5,6,7,8,9,10]. These effects are particularly critical for embankment dams founded on soft alluvial soils.

Previous studies have demonstrated that the core material plays a key role in the seismic behavior of earthfill dams [11,12,13]. Comparative investigations have shown that asphalt cores may provide improved deformation resistance compared to traditional clay cores under certain loading conditions [14]. However, the seismic response of asphalt-core and clay-core earthfill dams founded on soft alluvial deposits remains insufficiently explored, especially when subjected to different earthquake records.

This study presents a comparative numerical investigation into the influence of core material properties on the seismic performance of earthfill dams. Two geometrically identical embankment dams founded on soft alluvial soil—one incorporating an asphalt core and the other a clay core—are analyzed using PLAXIS2D (version 2023.2). The models are subjected to four earthquake records: Kobe, Kocaeli, Loma Prieta, and Northridge. The selected set of earthquake records was chosen to capture the effects of moderate to strong seismic events on the dynamic response of the model. The seismic response is evaluated in terms of crest horizontal acceleration and settlement, as well as lateral displacement of the downstream slope. The primary objective is to assess the effect of core material variability on dam behavior and to evaluate the feasibility of asphalt cores as a reliable alternative to traditional clay cores in seismically active regions.

2. Materials and Methods

2.1. Geometry of the Model

The numerical model of the earthfill dam consists of several layers, including the foundation, shell, filter, core, and drain. The core layer has a width of 1 m at the crest and 4 m at the base, while the filter layer has a uniform width of 1.5 m. As mentioned earlier, two main models were developed. All layers are identical in both models, with the only difference being the core material, which is modeled as clay in one case and asphalt in the other.

2.2. Boundaries and Dynamic Conditions

The numerical analyses in this study were carried out in two stages: static and dynamic. In the static analysis, equal degrees of freedom (DOF) were assigned to the lateral boundaries to ensure uniform deformation along the model sides. In the dynamic analysis, free-field boundary conditions representing drained soil behavior were employed to simulate far-field response, in accordance with recommendations from the literature [15,16,17]. To minimize the artificial reflection of seismic waves from the base of the model, a compliant base was considered. It was also fixed in both horizontal and vertical directions to prevent rigid-body motion. The lateral boundaries were restrained only in the horizontal direction, allowing vertical displacement. All the geometrical details of the numerical model are shown in Figure 1. The horizontal components of the selected earthquake records were applied at the model base as input motions. The time history as well as the maximum acceleration, a_max, of each earthquake is shown in Figure 2. Also, Table 1 presents the characteristics of the selected earthquakes (Arias Intensity and Predominant Period).

Figure 1. The model geometry.

Figure 2. Time history of the records applied at the base of the model.

Table 1. The characteristics of the selected records.

Earthquake	Arias Intensity (m/s)	Predominant Period (s)
Kobe	1.68	0.16
Kocaeli	1.32	1.4
Loma Prieta	1.35	0.22
Northridge	2.73	0.55

The earthfill dam model was set to plane strain, and the elements were set to 15-noded. The generated finite element mesh consisted of 21,960 elements and 41,739 nodes, and a fine mesh density was adopted for the analyses.

2.3. Asphalt Core vs. Clay Core

As mentioned earlier, to evaluate the influence of core material properties on the seismic behavior of earthfill dams, two numerical models were developed: one incorporating a clay core and the other an asphalt core. The clay core performs like a stress-dependent material in which stiffness and excess pore pressure generation significantly influence the response, while the asphalt core exhibits comparatively higher stiffness with time- and temperature-dependent viscoelastic characteristics. It should also be noted that clay may display viscoelastic behavior at low strain levels; therefore, the distinction between the two materials primarily lies in their stiffness and drainage-dependent response rather than a purely plastic versus viscoelastic classification.

2.4. Constitutive Soil Models

To model the different soil layers in PLAXIS 2D, the HS small constitutive model [18] was adopted. This model is widely used in geotechnical engineering to represent soil behavior under various loading conditions in numerical analyses. By incorporating both elastic and plastic responses, the HS small model enables accurate simulation of soil deformations, particularly at small strain levels. The model is defined by a set of stiffness, strength, and small-strain parameters. The definitions of these parameters and the values adopted in this study are summarized in Table 2.

Table 2. Material properties following HS Small as a soil model.

Input	Unit	Asphalt [19]	Clay	Filters	Drain	Foundation	Shells
${\gamma }_{sat}$	kN/m3	23.5	20.5	21	21	20	22
$E_{50}^{ref}$	MPa	300	40	80	100	60	180
$E_{oed}^{ ref}$	MPa	210	28	60	80	42	126
$E_{ur}^{ref}$	MPa	900	120	240	300	180	540
m	-	0.5	0.6	0.5	0.5	0.6	0.5
$G_0^{ref}$	MPa	500	60	150	200	80	300
${\gamma }_{0.7}$	-	5 × 10−3	1.5 × 10−3	5 × 10−4	3 × 10−4	10−3	5 × 10−4
c′ref	kPa	50	20	0	0	5	5
${\phi }^{\prime }$	°	18	22	36	40	32	38
$\Psi$	°	5	0	5	7	2	5
Kx, Ky	m/s	10−10	10−8	10−4	5 × 10−2, 10−2	10−4	10−3

${\gamma }_{sat}$: Saturated unit weight; ${\mathrm{E}}_{50}^{\mathrm{r}\mathrm{e}\mathrm{f}}$: Secant stiffness in the standard drained triaxial test; ${\mathrm{E}}_{\mathrm{o}\mathrm{e}\mathrm{d}}^{\mathrm{r}\mathrm{e}\mathrm{f}}$: Tangent stiffness for primary oedometer loading; ${\mathrm{E}}_{\mathrm{u}\mathrm{r}}^{\mathrm{r}\mathrm{e}\mathrm{f}}$: Unloading/reloading stiffness; m: Power for stress-level dependency of stiffness; ${\mathrm{c}}_{\mathrm{r}\mathrm{e}\mathrm{f}}^{\prime }$: Cohesion; ${\phi }^{\prime }$: Friction angle; ${\gamma }_{0.7}$: Shear strain at which G_s = 0.722 G₀; ${\mathrm{G}}_0^{\mathrm{r}\mathrm{e}\mathrm{f}}$: Shear modulus at very small strains; $\Psi$: Angle of dilatancy.

3. Results and Discussion

In this section, the results of the parametric analyses are presented graphically and discussed in detail. The results include horizontal accelerations calculated on the crest and downstream slope, lateral displacement on the downstream slope, and settlement on the crest.

3.1. Horizontal Acceleration Recorded on the Crest

Following the dynamic analyses of two earthfill dam models with clay and asphalt cores subjected to four earthquakes with different amplitudes, the results are plotted and presented in Figure 3. As shown in the figure, the maximum crest acceleration is 0.29 g for the Northridge earthquake, which has the highest input amplitude. The corresponding crest accelerations are 0.17 g, 0.02 g, and 0.022 g for the Kobe, Kocaeli, and Loma Prieta earthquakes, respectively. The results indicate that the models with clay and asphalt cores exhibit nearly identical crest acceleration responses. This confirms that variation in core material has a negligible influence on the transmission of acceleration from the foundation to the crest, as the global dynamic response is primarily governed by the dam geometry and the stiffness and damping characteristics of the shell and foundation materials rather than the core.

Figure 3. Acceleration time-history responses at the crest for different core materials under various earthquake records: (a) Kobe, (b) Kocaeli, (c) Loma Prieta, and (d) Northridge.

3.2. Horizontal Acceleration Recorded on the Downstream Slope

Figure 4 presents the horizontal acceleration time histories recorded at the downstream slope. As shown, the maximum acceleration reaches 0.29 g for the Northridge earthquake, which has the highest input amplitude. The corresponding accelerations are 0.15 g, 0.019 g, and 0.02 g for the Kobe, Kocaeli, and Loma Prieta earthquakes, respectively. The results indicate that the accelerations transmitted to the downstream slope are slightly lower and occur more gradually than those at the crest. This behavior can be attributed to the higher potential for lateral deformation of the downstream slope, which allows part of the seismic energy to be dissipated through slope movement, thereby reducing dynamic amplification and limiting the transferred acceleration. Again, the results indicate that the models with clay and asphalt cores show similar crest acceleration responses.

Figure 4. Acceleration time-history responses on the downstream slope for different core materials under various earthquake records: (a) Kobe, (b) Kocaeli, (c) Loma Prieta, and (d) Northridge.

3.3. Settlement Recorded on the Crest

Figure 5 illustrates the crest settlement for different seismic scenarios. It is evident that the models with a clay core exhibit noticeably greater crest settlement than those with an asphalt core. This behavior is primarily attributed to the lower stiffness and higher compressibility of the clay core, which allows larger cyclic strains and volumetric contraction under seismic loading, whereas the asphalt core, due to its higher stiffness and viscoelastic behavior, provides greater resistance to deformation.

Figure 5. Settlement time history responses calculated at the crest for different core materials under various earthquake records: (a) Kobe, (b) Kocaeli, (c) Loma Prieta, and (d) Northridge.

For the Kobe earthquake, the maximum crest settlement is 0.002 m and 0.01 m for the asphalt and clay cores, respectively. For the Kocaeli earthquake, the corresponding settlements are 0.0004 m and 0.0006 m, respectively. Similar values are observed for the Loma Prieta earthquake, with maximum settlements of 0.0004 m and 0.0006 m for the asphalt and clay cores, respectively. For the Northridge earthquake, the maximum crest settlement reaches 0.02 m for the asphalt core and 0.028 m for the clay core.

3.4. Lateral Displacement Recorded on the Downstream Slope

With respect to the lateral displacement of the downstream slope, Figure 6 again shows that variations in core material properties have no significant influence on the seismic response of the earthfill dam. This is because the lateral deformation of the downstream slope is mainly governed by the dynamic behavior of the shell materials and overall dam geometry, while the core primarily controls seepage rather than slope deformation under seismic loading. The maximum lateral displacement, equal to 0.097 m, is observed for the Kobe earthquake, whereas the minimum value of 0.02 m occurs under the Loma Prieta earthquake.

Figure 6. Lateral displacement time histories calculated on the downstream slope for different core materials under various earthquake records: (a) Kobe, (b) Kocaeli, (c) Loma Prieta, and (d) Northridge.

4. Conclusions

• The parametric dynamic analyses of earthfill dams with clay and asphalt cores indicate that the dam with a clay core experiences greater crest settlement under cyclic loading compared to the dam with an asphalt core. This behavior is primarily attributed to the elastoplastic nature of the clay core, which, due to its lower stiffness and higher compressibility, undergoes larger deformations during seismic loading.
• In earthfill dams founded on alluvial soils, cyclic shear loading generates excess pore water pressure and reduces effective stress, leading to permanent deformations. Under these conditions, the clay core—because of its lower stiffness and higher compressibility—shows larger cyclic strains and volumetric contraction, resulting in greater crest settlement. In contrast, the asphalt core, with higher stiffness and viscoelastic damping, limits strain accumulation and redistributes stresses more effectively, thereby reducing the development of permanent settlement of the crest during seismic events.
• The similar lateral displacements observed for both core types are primarily governed by the response of the alluvial foundation. Alluvial soils are inherently susceptible to lateral deformation under cyclic loading [20], and their influence is manifested through two main mechanisms. First, their low stiffness: the deformation modulus of the alluvial foundation is considerably lower than that of the embankment and core materials, leading to increased dam displacements and a higher potential for sliding during seismic excitation [21]. Second, seismic amplification: soft alluvial soils amplify the input ground motion, thereby increasing the dynamic demand transmitted to the dam body [22].
• It was observed that the highest accelerations at both the crest and downstream slope occurred for the model subjected to the Northridge earthquake, as it has the strongest input motion with the highest peak ground acceleration and energy content among the considered records.
• It was observed that the maximum crest settlement occurred for the model subjected to the Northridge earthquake, as its higher intensity and energy content induced larger cyclic strains and permanent deformations within the dam body.
• It was also observed that the Kobe earthquake induced larger lateral displacements than the Northridge earthquake, despite having lower peak accelerations, due to its longer effective duration and frequency content being closer to the fundamental period of the dam–foundation system, which promoted greater cumulative deformation.
• Under extreme conditions, such as regions with high seismic activity, heavy rainfall, and short construction seasons, asphalt cores are more suitable due to their ease of implementation and more efficient compaction compared to clay cores.