A field campaign was essential for material identification using the Maignien (1980) method. Three main soil types were identified: lateritic gravelly soils, fine sandy soils with varying clay content, and fine clayey soils.

Regardless of the intended use of the soil, it is crucial to understand its nature, composition, and particle size distribution ( Manefouet, 2016 ; Boutahir, 2018 ). A total of thirty disturbed soil samples were collected using manual auger drilling (targeted sampling strategy—STM) and transported to the laboratory for physical characterization.

The natural water content of the studied soils was determined using the successive weighing method before and after drying at 105˚C, in accordance with standard NF P 94-050 (1995) . The specific gravity of the soil grains was determined using a pycnometer, following standard NF P 94–054 (1991) . The consistency of the soils was assessed based on the Atterberg limits. The liquid limit (ωL) was determined using the Casagrande apparatus, and the plastic limit (ωP) was obtained using the rolling method. These Atterberg limits were measured following standard NF P 94-051 (1993) , and the plasticity index (IP) was calculated as the difference between ωL and ωP. Particle size distribution was performed by dry sieving after washing, according to the procedures of standard NF P 94-056 (1996) . The void ratio and porosity were calculated using established relationships among the various physical characteristics of the soils.

In this study, a light dynamic penetrometer with constant energy was used as the geotechnical investigation tool, in compliance with EN ISO 22476-2 (2005) . The light dynamic penetrometer test is a soil investigation method that evaluates the terrain and provides a characteristic soil parameter ( Afnor, 1990 ). This test allows a qualitative assessment of the resistance of the traversed layers.

The dynamic resistance to failure Qd (kg/cm²) at the penetrometer tip is expressed using the Dutch formula:

$Qd=\frac{M^2\times H\times g}{A\times e}\times \frac{1}{M+M^{\prime }}$ (1)

$e=h/N$ (2)

where:

$Qd$: tip resistance (MPa or kg/cm²);

M: hammer mass (kg);

H: drop height of the hammer (m);

g: gravitational acceleration (m/s²);

M′: mass of the struck components (kg);

A: cross-sectional area of the tip (m²);

e: penetration per blow (m);

N: number of blows;

h: total penetration depth (m).

The allowable stress σ all of the foundation soils under the Service ability Limit State (SLS) was specifically assessed using the following formula ( Sanglerat, 1965 ):

${\sigma }_{adm}=\frac{Q_d}{s}$ (3)

where s is the safety factor.

The safety factor used to derive allowable stress from dynamic resistance is 20, which incorporates a global safety factor of 3 and a reduction coefficient of 6.6 ( Zagalo et al., 2017 ).

The geotechnical parameters measured both in situ and in the laboratory display substantial variability, which can lead to uncertainties in selecting representative values and in the accuracy of final geotechnical calculations ( Saporta & Niang, 2003 ). This approach relies on examining the relationships between the measured parameters in order to highlight the influence of physical properties on the dynamic resistance of soils—a critical factor in structural design.

PCA graphically displays the differences and/or similarities among samples subjected to selected analytical conditions ( Cordella, 2010 ; Annad & Lefkir, 2022 ). Samples with similar characteristics cluster together to form more or less distinct aggregates. In this work, the XLSTAT 2018 software was used to generate graphical representations and data matrices.

Overall, the physical parameters studied across the various soil types show considerable variability. According to the GTR (2000) classification system, the soils are subdivided into three subcategories: fine clayey soils (A3), silty-sandy fine soils (A2), and clayey or silty lateritic gravel soils (B6).

The natural water content ranges between 12.61% (on gravels materials) and 53.13% (on fines materials), with an average of 29.92%. The standard deviation and the coefficient of variation are 8.7 and 0.29, respectively. This variation in the water content of the samples can be explained by the role of the fine fraction in retaining water in the soil, and by the influence of the environment. Indeed, materials with low water content are proportionally coarser in grain size. An average value for natural water content in fine clayey soils is lower than the 48% reported by Ananfouet Djeufack (2012) in fine basaltic soils of Dschang and the 59.95% observed in the fine soils of Mouraye, Chad by Zagalo et al. (2017) . However, it is comparable to the values obtained by Ngapgue et al. (2008) in the fine soils of Bafoussam. The low water content (14.25%) is observed on some gravelly materials (ES03, ES08, ES11, ES17, ES20 and ES23) and for others, it is higher, at 31.87% (ES18, ES19, ES21 and ES29). The value of 14.25% is close to the 12.50% obtained by Aboubakar et al. (2025) on gravelly soils in Ngaoundal. These observations show that climatic and morphological conditions could also explain these variations in water content, with average annual precipitation of 1782.3 mm/year and altitudes of up to 1800 m.

The specific gravity ρs of the materials ranges from 2.53 to 3.21 g/cm³, with an average value of 2.81 g/cm³. The data dispersion appears low, with a coefficient of variation of 0.06. The specific gravity values (2.53 - 3.21 g/cm³) indicate that these materials have good to excellent performance in road construction ( Nwaiwu et al., 2006 ; Paige-Green et al., 2015 ), except for the samples (ES02, ES05, ES07 and ES10) with values <2.60 g/cm³. This average value is close to those obtained by Onana et al. (2017) (2.69 and 2.81g/cm³) and Nzabakurikiza et al. (2016) (average ρs = 2.81 g/cm³) on gravelly materials. The high specific weight values would be linked to the pre-presence of ferromagnesian minerals and silica as observed by Manefouet (2016) on soils in the area. This observation confirms the findings of Nzabakurikiza et al. (2016) , who believe that a high specific gravity ρs value is due to the presence of iron. Since void ratio is directly related to the porosity, any increase in one will lead to an increase in the other. Consequently, the porosity of these soils varies from 28 to 58, with an average of 44, indicating that these soils are moderately porous overall. The standard deviation and the coefficient of variation are 0.08 and 0.18, respectively. The void ratio value is significantly lower (0.82) than the 1.38 reported by Zagalo et al. (2017) , and the values ranging from 7 to 14 observed by Rutledge (1944) , which represent extreme cases of materials with very high void ratios. These findings suggest that the soils analyzed in this study may exhibit relatively low compressibility.

With respect to particle size distribution, gravels, sands, and particles smaller than 0.08 mm (silts and clays) are present in the samples ( Figure 2(a) ). Gravel content varies between 0% and 72.94%, with an average of 24.06%; sand content ranges from 1.73% to 44.46% (mean: 21.62%); and fine particles range from 3.51% to 98.5%, with an average of 52.28%. The high proportion of coarse elements in the materials studied testifies to their extensive lateritization, with the presence of cuirasses in places. This percentage-age is close to the 74% obtained by Manefouet (2016) on gravelly soils developed on basalt, indicating that these soils were formed under the same geo-morphological conditions as those studied.

The Atterberg limits reveal liquid limit values (ωL) ranging from 44.4% to 85.8% (mean: 63.01%), plastic limit values (ωP) from 26.11% to 55.71% (mean: 40.96%), and plasticity index values (IP) ranging from 10.27% to 36.13%, with an average of 22.25%. The plotting of IP and ωL values on the Casagrande plasticity chart (1948) indicates that the analyzed soil samples consist mainly of plastic to highly plastic silts, with the exception of samples ES11 (low plasticity clay) and ES19 (low plasticity silt) ( Figure 2(b) ). The studied materials are clayey and highly plastic materials according to Casagrande plasticity chart. This high degree of plasticity and clayiness is due to the high kaolinite content.

A total of ninety (90) dynamic penetrometer test points were carried out at varying depths, reaching up to 6.40 meters. The resulting penetrometer logs ( Figure 3 ) show the relatively contrasting mechanical behavior of the subsurface layers. Analysis of the penetrograms indicates that in fine soils, dynamic resistance values fluctuate between 0.2 and 1.5 MPa ( Figures 3(a)-(c) ); in fine sandy soils, they range from 0.2 to over 3 MPa ( Figures 3(d)-(f) ); and in lateritic gravelly soils, values are higher, ranging from 1 MPa to more than 11 MPa ( Figure 3(g) and Figure 3(h) ), and even higher in ferricrete horizons, where refusal was observed ( Figure 3(i) ).

The lithological profile ( Table 2 ), reveals a sequence of 3 to 5 distinct layers with various thinknesses. Overall, the allowable bearing pressures (σ_all) across the different soil layers range from 0.12 bar to more than 5.84 bars, depending on depth. In clayey soils, allowable pressures vary between 0.12 and 1.07 bars. These values increase to between 0.77 and 2.3 bars in silty-sandy fine soils, and from 2.5 to 5.2 bars in lateritic gravelly soils, exceeding 5.84 bars in cuirass horizons. This increase in allowable bearing pressure may be attributed to several factors, including grain size distribution, soil compaction, and porosity ( Dysli & Bobelli, 2001 ; Martin, 2008 ). These findings on resistance differences based on granulometry are consistent with those reported by Demanou Messe et al. (2022) for similar lateritic soil types.

The eigenvalue table for the extracted factors reveals two dominant factors among the nine identified. The factors selected must together account for at least 75% of the total variance. Factor 1 (F1) accounts for 57.87% and Factor 2 (F2) contributes 18.97% ( Table 3 ), which justifies the retention of these two principal axes for analysis, as they represent more than half of the total explained variance. The remaining seven factors are considered residual.

The contribution of the variables to each factor helps identify which parameters characterize each component. It is observed that water content (ω), fine fraction (%fines), porosity (n), and void ratio (e) contribute positively and predominantly to Factor 1 ( Table 4 ). On the other hand, allowable stress (σall) and specific gravity (ρₛ) contribute most significantly—but inversely—to Factor 1. This indicates that Factor 1 is characterized by parameters related to the soil’s hydric state. The liquid limit (ωL), plastic limit (ωP), and plasticity index (IP) are weakly correlated with Factor 1, but contribute predominantly to Factor 2—with more than 80% influence—compared to other parameters (which are under 30%).

The combined analysis of physical parameters and allowable stress ( Table 5 ) shows strong correlations (R > 0.6) between σ_all and variables such as water content (r = −0.67), specific gravity (r = 0.66), porosity (r = −0.68), fine fraction percentage (r = −0.70), and void ratio e (r = −0.65). These results are consistent with the findings of Zagalo et al. (2022) . However, weak correlations (R < 0.4) are found with Atterberg limits (ωL, ωP) and IP with r equal −0.30, −0.06 and −0.39 respectively, suggesting that plasticity has minimal and negative influence on the mechanical behavior of these soils. These observations on the weak influence of Atterberg parameters on dynamic strength are similar to those made by Zagalo et al. (2022) on Mouraye soils (ωL= 0.08, ωP = 0.13 and IP = 0.03). However, a material with a high Ip may be subject to significant volumetric variations linked to its clayey nature ( Nadir, 2024 ). Consistency parameters are therefore essentially linked to the mineralogical composition, granularity, clay content, humidity and cohesion of the material. The materials studied are essentially plastic silty according to Casagrande’s (1948) abacus, which could explain the weak relationship observed between clay-related consistency and dynamic strength. The variable correlation circle ( Figure 4(a) ) highlights their significance: Group 1 (σ_all and ρs) lies opposite Group 2 (water content, fine fraction, and porosity). The relationships between these parameters and Factors F1 and F2 indicate that F1 likely represents the “material nature” of the soils, while F2 reflects “consistency”.

Once the factors are identified, their meaning is clarified by projecting the variables and individual samples onto the factorial axes and analyzing their distribution ( Figure 4(b) ). The sample projection reveals three distinct groups. Lot 1 mainly consists of gravelly soils, particularly highly plastic silty gravels (ES3, ES8, ES11, ES17, ES18, ES19, ES20, ES21, ES23, ES29); Lot 2 includes silty fine sands (ES01, ES04, ES06, ES09, ES12, ES13, ES14, ES15, ES16, ES30); and Lot 3 consists of highly plastic silty clays (ES02, ES05, ES07, ES10, ES22, ES24, ES25, ES26, ES27, ES28).

The mechanical behavior of the soils appears particularly sensitive to variations in physical properties, especially water content and fine particle percentage. A classification of the soils based on strength is established using the Robertson (1990) soil classification chart, which is based on cone tip resistance (kg/cm²) as a function of sleeve friction ( Jacquard & Boutet, 2016 ).

The modification of this chart into a diagram plotting dynamic resistance against fine fraction percentage ( Figure 5 ), which influences rod penetration, reveals a distribution of sample points. The diagram produces a soil classification based on these parameters (dynamic resistance Qd – %Fines/Qd ratio). A decreasing trend in dynamic resistance illustrates the transition of materials from sands and gravelly clays to silts and clay-rich soils. This confirms the influence of physical parameters such as fine particle content on the dynamic resistance of soils. An increase in fine content—particularly clays with high water absorption capacity—leads to a decrease in the material’s dynamic resistance. This observation is in line with that made by Jurado Sastre et al. (2018) on the considerable influence of the proportion of fine fraction on dynamic resistance. Indeed, the high dynamic resistances observed progressively decrease from lateritic gravels to clays.

Neglecting the physical properties of soils in engineering design can lead to numerous consequences. Parameters such as water content, particle size distribution, and porosity may be key factors in the degradation of structures, as reported by Jusoh & Osman (2017) . Settlement and landslides are examples of structural pathologies directly linked to soil characteristics and their physical properties. In addition, a reduction in the water content of a material considerably improves its mechanical properties, such as dynamic strength and dry density ( Dysli & Bombeli, 2001 ). The mechanical strength of a material depends essentially on its solid matrix, which is a function of its granularity. From these analyses, it emerges that soils with a high-water content, a considerable proportion of particles, high porosity and high plasticity have poor mechanical strength and could be prone to excessive settlement.

Furthermore, a lack of understanding of subsurface property variability can result in short- and medium-term additional costs during project implementation. These may arise due to unforeseen hazards or structural damage stemming from inadequate adaptation of engineering designs to subsurface constraints ( Guerrero, 2014 ; Sivelle, 2016 ).