Geotechnical Interpretation of the Geological Structure of Loess Covers in Lublin Region

An essential part of any building structure is the foundation, which is a critical zone of contact between the engineering structure and the ground. This place links the “work of man” made in a controlled and regulated manner with the “work of nature”, which was created over millions of years without human control and presence, and has often been transformed in later periods. In the case of buildings, we can influence the process of their construction and choose solutions and materials with appropriate parameters, we get the subsoil as it was created and we need to adapt to it. The spatial variability of the subsoil and complex non-linear behaviour introduces a lot of uncertainty to the design process.

Geology studies the structure of the subsoil, its history, and its creation processes, while the proper design of the building structure, including the foundations, is the role of the designer of the structure. A connecting link between these two fields is geotechnical engineering, which in a way assumes the role of a translator of geological processes into numerical values needed in the design process. Awareness of geological processes that occurred during the formation of the subsoil helps to better derive parameters and understand their meaning. To correctly design a building, it is essential to properly interpret geological conditions, and this requires knowledge of both geology and construction.

The identification of subsoil layers includes both geological and geotechnical works. The scope of geological testing should also cover the identification of geological genesis and hazards, while tests in the field of geotechnical engineering are primarily aimed at defining parameters that describe subsoil behaviour.

The paper presents the author's experience and remarks related to the identification and evaluation of the behaviour of the subsoil from the Lublin region in geotechnical terms.

A characteristic feature of the geological structure of the Lublin region is loess covers. Although loess soils occupy a relatively small area of Poland (an estimated 7% [1]), a significant part of this loess cover occurs in the Lublin region, mainly on the Nałęczów Plateau (Fig. 1). The thickness of loess covers ranges from several to even 20 meters, which makes them the main subsoil for setting building structures in the Lublin region [2]. Proper geotechnical interpretation and determining appropriate parameters that describe the behaviour of these specific geological deposits are crucial for the proper setting of buildings. In the literature, you can find examples and guidelines from research work carried out using various techniques [3,4,5,6,7,8].

Figure 1.

The main factor in creating loess was the wind, but often other processes occurring simultaneously or in a later period also had a significant impact on its parameters. The genetic group of loess includes both deposits of “pure” aeolian origin, as well as loess-like deposits which were created with the participation of other factors or were partially transformed in the later period. For engineering purposes in the Lublin region, three main facies groups should be distinguished: the aeolian (typical loess), the aeolian-diluvial, and the aeolian-alluvial [2, 10, 11]. The structure of the loess cover in the Lublin region is basically repeatable, which is shown in the schematic representation in Fig. 2.

Figure 2.

The upper, surface zone, 1–2 m on average, consists of aeolian-diluvial loess. These are the soils that were originally accumulated by the wind in the form of typical loess and then were re-deposited or transformed by flora and fauna, so they are usually darker in colour, brown or dark beige. There are also numerous streaks and thin silty-clayey stratifications. In terms of granulometry, these deposits are usually classified as clayey silt to silt. They have a higher humidity and a hard-plastic or plastic consistency.

The main subsoil is formed by typical loess from the aeolian facies. Its thickness reaches up to 10–15 meters. In terms of granulometry, these are silts, macroscopically homogeneous, sporadically occurring with thin sandy interbeddings. They are usually light beige or yellow in colour, with low wetness and compact or hard-plastic consistency. Sometimes in the deeper parts of the subsoil, older aeolian loess can also be distinguished, with a high content of sandy fractions.

Loess soils found at the greatest depths, below typical loess, are those which belong to the aeolian-alluvial group and rest directly on detrital bedrock or water-glacial sands. This facies is much more heterogeneous and occurs in the form of clayey silts or silts with interbeddings of sand and silty sands. Their high variability is also reflected by the parameters measured in in situ tests.

Loess are considered to be specific soils and are mainly associated with the phenomenon of collapsibility, which is associated with macroporosity [12,13,14]. It is also important that most of the loess subsoil, in particular silty “typical loess”, occurs in the unsaturated state. High porosity in connection with the unsaturated state is the reason why most risks associated with these soils are related to the impact of water. This applies to both static and dynamic impacts. Collapsibility is a result of pore saturation and this concerns macroporous soils. However, an equally dangerous and more common phenomenon is sufosion, associated with the washing of soil particles as a result of the dynamic impact of water. In addition, due to the action of water, loess undergoes rapid plasticisation and is susceptible to the phenomenon of thixotropy. Another threat associated with loess is landslides, which often occur violently, in the form of shear surfaces that are close to vertical.

When designing a building structure, one needs to analyse both soil strength by checking the load capacity of the subsoil and its deformability by estimating the settling of the subsoil caused by the loads transmitted by the foundation. In both cases, it is important to adopt a proper geotechnical subsoil model, which is developed on the basis of geological structure. In the process of building a geotechnical model, two main elements are distinguished: the division into geotechnical layers and deriving representative parameters that describe the behaviour of the soil which is included in a given layer.

In many cases, especially for simple structures, geotechnical models can be identical to geological models, but often these models should differ. Moreover, for the same subsoil conditions, geotechnical models may vary depending on the type of structure analysed, its dimensions, the method and depth of the foundation, and the loads transferred. This applies both to the division into layers and the derived parameters. Especially in the case of complex building structures which cause the interaction between significant volumes of subsoil, geotechnical models should be developed taking into account more factors than just geological layers. However, the knowledge of geological conditions is extremely important in this process.

In geotechnical and geological studies, ground layers are identified taking into account the following four factors: stratigraphy, genesis, lithology and parameters. At that last stage, the so-called “leading parameter” is usually assigned to layers. In Poland, it is most often the density index I_D or liquidity index I_L. While adopting the density index is generally accepted in the geotechnical environment, the validity of using the liquidity index is debatable, especially for low-cohesive soils or when the soil layers in the same state occur at different depths and sometimes are additionally genetically different. Soil stiffness and its bearing capacity depend not only on the state, i.e. wetness but also on the current state of stress and strain, granular soil structure, the state of saturation, load history and load velocity. Therefore, the construction of geotechnical models that identify layers based only on lithology and subsoil state sometimes leads to significant and excessive simplification, especially when these layers occur at different depths. In geotechnical identification, in situ tests play a very important role, as they allow to conduct high-frequency measurements of subsoil parameters. On the basis of drilling and a superficial macroscopic assessment only, even with the verification of individual samples with laboratory tests, it is difficult to properly identify the necessary properties of the soil.

The ground is a typical medium with a strongly nonlinear load-deformation behaviour. In addition, the course of deformation is affected by several other factors, such as the state of stress, degree of saturation, load velocity, time, degree of pre-consolidation, etc. [15, 16]. Soil stiffness should be described by a function whose graphical representation is called the stiffness degradation curve [15]. Despite this knowledge, it has been accepted in the engineering practice to use generalised parameters describing deformability – reduced to a single value, intended to reflect secant moduli - for the simplification of calculations. In most cases, the error resulting from such simplification is acceptable as long as the moduli have been determined in the range of relevant strains and stresses corresponding to the given design project. It is assumed that for most structures working in the “typical range of strain and stress”, the most appropriate is the elastic modulus corresponding to the deformation produced at 50% soil-bearing capacity marked as E₅₀ [17]. Subsoil behaviour is very well represented by triaxial tests. The results of these tests perfectly illustrate the non-linearity of the load-strain behaviour as well as the increase in stiffness with the increase in stresses (Fig. 3). However, it should be borne in mind that the classic constrained moduli determined in oedometer tests are a significant simplification and describe the subsoil in the range of large deformations. They are not suitable for all soils and design projects. As a rule, they are better for representing the behaviour of weaker bearing soils, while these parameters are not recommended for estimation of settling of loess soils, as the obtained values are usually significantly overestimated [18].

Figure 3.

Given the above, an accurate description of geotechnical layers requires considering the actual state of stress in the subsoil. Thus, even when soils are lithologically similar, they have similar humidity (and state), and even cone resistance q_c, as long as layers lie at significantly different depths, they should be described with different values of elastic (or constrained) moduli. Fig. 3 and Fig. 4 show the interpretation of elastic moduli taking into account the state of stress in the subsoil [15, 19].

Figure 4.

Even if perfect subsoil homogeneity is assumed, the ground directly below the foundation works in the range of larger deformations from the deeper subsoil, and is also less stiff. With depth, soil stiffness increases, and deformations resulting from the applied load decrease. Therefore, to properly model soil stiffness, secant moduli with higher values should be used as depth increases. A certain representation of this is the triaxial tests performed on several samples at different initial effective stresses that correspond to vertical in situ stresses. Admittedly, this is not a completely accurate representation, because triaxial tests assume an isotropic state of initial stress, and in fact, in normally consolidated soils, the horizontal stresses are less than the vertical [19], but for engineering purposes, this simplification is considered acceptable. Another issue is that as the depth from the foundation increases, the influence of the parameters of a given layer on the settling of the building is reduced. Therefore, simplifications and errors at greater depths are less important for the design result. Thus, in the construction of geotechnical models, more importance should be attached to the layers in the direct area of the foundation and they should be identified in more detail. As depth increases, layers should be classified in a more generalised way.

A very good method to identify the structure of the subsoil is in situ tests, mainly those providing quasi-continuous data such as CPT/CPTU static sounding and dilatometric tests. In DMT studies, it is possible to determine deformability in a natural stress state. If seismic measurements (e.g. SDMT or SCPTU) are additionally performed during these tests, the initial shear modulus G0 can also be determined based on the measurement of the velocity of wave propagation in the ground.

An important issue in the case of loess soils is a large variation in stiffness, with relative macroscopic uniformity. Typical loess in the Lublin region is usually in the unsaturated state and solid state (I_L<0), while CPT studies indicate that cone resistance q_c for loess with I_L<0 ranges widely from 4 to 12 MPa, and constrained moduli from dilatometric tests M_DMT from 20 to 90 MPa [20, 21]. Such a wide range of values indicates a strong differentiation of the loess subsoil and the need to identify geotechnical layers not only in terms of I_L, but also, and even above all, in terms of cone resistance q_c. This parameter reliably reflects soil stiffness, and at the same time, it is a statistically large set of data. CPT test data are the basis for deriving geotechnical parameters by correlation, but this requires the use of formulas developed or proven for local subsoil conditions.

Typical loess most often occurs in a solid to hard-plastic state. Plasticised typical loess is much less common and it is usually found only in the upper and bottom layer of loess cover. An example of a subsoil composed of typical loess soils is a fragment of the geotechnical cross-section in Fig. 4. As in Poland geological layers are usually identified using I_L as the leading parameter, this results in the entire loess subsoil being included in one geotechnical layer. However, as in situ tests indicate, both diverse cone resistances q_c and constrained moduli are found in loess, which allows for distinguishing separate sublayers. In this example, sublayers were identified using description with a representative q_c.

Using parameter I_L as the leading one is related to the fact that PN standards, including PN-81 B-03020, have been in use in Poland for a long time [22]. Globally, however, this parameter is given much less importance [23]. The [22] standard used the liquidity index to derive strength and deformation parameters by correlation. In spite of the option to determine geotechnical parameters using correlation, in Polish practice, this was most often done incorrectly because in most cases I_L was determined on the basis of macroscopic analysis, and occasionally with individual laboratory tests. This is an erroneous approach because in the related standard PN 88 B-04481 [24] there was a provision saying that “the above method (thread rolling test – author's note) should not be used to determine values of the liquidity index I_L, and only to determine the state of the ground”. This provision is regularly breached in engineering practice for all soils. Moreover, for loess soils (silts in granulometric classification), the difference of one thread caused a change of state, so even with such an approach it is impossible to determine the liquidity index with an accuracy greater than 0.25. However, another provision: “I_L should not be used for low-cohesive soils” pointed out that silts should not be classified using I_L at all, and that this value should certainly not be used to determine significant strength and deformation parameters by correlation. Practically no one follows this principle, and most people are not even aware of its existence.

Due to the preference for the I_L parameter in Poland and the expectations of designers to determine its value in the documentation, the author performed a number of calibration tests between q_c and the I_L value determined in the laboratory. The first, initially determined correlations were described in [25], while more extensive studies in which the formulas were updated were conducted in 2022 in the Nałęczów area, located in the centre of the Nałęczów Plateau. Selected results are presented in Fig. 5. In this case, I_L values match the trend shown by q_c, but the variability of these values should be taken into account. In cases like the one described above, when data from wide-ranging laboratory tests are available, extreme results can be rejected and averaged values can be adopted as a basis for concluding. However, in practice, verification tests are carried out only on selected individual samples, taken e.g. every 2 m, and thus they cannot be representative of a larger range. Macroscopic examination does not allow for determining the correct I_L value, but only – approximately – the state. In addition, errors occur in storage, transport of samples, and laboratory tests. With this approach, the impact of local variability and the researcher's subjectivity in the macroscopic assessment cannot be eliminated.

Figure 5.

To sum up the above, according to the author, the q_c value should be the basis for geotechnical identification, as it is statistically more valuable, and also reflects soil strength and its stiffness, and indirectly (but not always) state. In the author's opinion, it is possible to use the I_L value to describe loess soils, but it should only be treated as an auxiliary parameter, of secondary or tertiary importance. Calibration tests made it possible to develop interpretation patterns of the transition from q_c to I_L, but these have certain limitations and should be used under appropriate conditions; selected probings should be verified by drilling a borehole in the immediate vicinity. A representative example of a geotechnical cross-section of the loess subsoil in the Lublin region, which was constructed based on in-situ tests verified by drilling boreholes, is shown in Fig. 6.

Figure 6.

Although loess soils are associated with many adverse phenomena, those found in the Lublin region generally constitute a favourable subsoil for foundation setting. However, they should be adequately protected from water, and ground works should be carried out in an appropriate regime. Geotechnical parameters should not be determined based on correlation with liquidity index I_L; much better consistency of results is achieved with correlations with cone resistance q_c. Numerous studies of the author [10, 21] have allowed him to derive correlations for constrained moduli M_CPT in relation to M_DMT moduli, which provide reliable data in the analysis of soil settlement, as well as of simplified determination of shear modulus G₀. About pressuremeter tests, initial correlations were also derived between cone resistance, pressuremeter modulus E_M, and limit stresses p_L.

Based on the cone resistance q_c, a generalised assessment of the suitability of loess subsoil for foundation setting can be made. Factors such as the type of structure, volume and type of loads, depth of foundation, etc., must always be taken into account, but generalising, for typical foundations (e.g. footings with sizes up to 3 m, carrying stresses up to 300 kPa) it can be assumed that favourable and very favourable conditions for foundation setting are provided by loess soils with q_c > 4.0 MPa. Loess soils with q_c < 2.5 MPa should be considered unfavourable, and those with q_c < 1.2 MPa – very unfavourable. It is also necessary to pay attention to areas with q_c < 3.0 MPa, because when they are solid or hard-plastic soils, low cone resistance is the result of the high porosity of the soil, so it can be an area susceptible to collapsibility. The cumulative statistical distribution of cone resistances q_c recorded in the studies of loess in Lublin together with the assessment of its suitability for foundation setting is presented in Fig. 7.

Figure 7.

The paper presents the author's position on the geotechnical interpretation of loess subsoil, which consists of the construction of a geotechnical model. Simplifications in calculation models in relation to actual conditions are inevitable and even desirable. This applies both to the range of layers and the parameters. However, the level of simplification should be adapted to the design project. For small building structures, there is no need for advanced analysis, while for larger and more complex structures, a more detailed description of subsoil behaviour should be required, which necessitates more tests. In simplified terms, it can be assumed that for the first geotechnical category, the geotechnical model is the same as the geological model. In the case of higher geotechnical categories, more complex structure of the subsoil and the building, as well as higher load values, more detailed analysis is required, as well as the development of a geotechnical model, which may or even should differ from the geological model.

A good basis for the geotechnical assessment of the subsoil is in situ tests. In particular, studies providing a large amount of data, such as CPTU and DMT, allow for the development of a suitable model of the subsoil, as well as the assessment of its parameters in statistical terms. They are also the basis for selecting the sampling sites to determine specific geotechnical parameters. Ultimately, the following conclusions were drawn:

The basis for the identification of geotechnical layers should be cone resistance q_c.