Review of classification methods to determine expansion degree of expansive soils in Indonesia

Expansive soils are one of the problematic soils that can cause damage to structures, such as buildings and roads. In Indonesia, the geographical distribution of expansive soils is quite widespread, in many regions including Java, Sumatera, and Papua. This study was conducted by reviewing previous research on the expansion degree of expansive soils. The suitability of the expansion degree indicated by the activity method, Van der Merwe’s criteria, and Chen’s criteria were evaluated by comparing a series of laboratory test results to these criteria. It was observed that the expansion degree indicated by these criteria did not necessarily reflect the actual swelling potential. Other indicators, other than typical index properties, are needed to predict the expansion degree more accurately.


Introduction
Expansive soils are one of the problematic soils that have significant swelling and shrinking behaviour due to changes in water content. When there is heavy rainfall, the water content in the soil increases and causes an increase in volume. However, when the water content decreases, there will be soil shrinkage because of the reduced volume of the soil. The existence of expansive soils in Indonesia is geographically distributed over several regions of Indonesia, including regions in Java and Sumatera in the west to Papua in the east. Expansive soils often cause damages to buildings and other infrastructures, such as reduced foundation bearing capacity, longitudinal cracks in road pavements, heaving of soil causing bumpy roads, and instability due to reduced shear strength of soils.
From the soil formation processes viewpoint, a soil is the result of physical and chemical weathering processes which consequently affect its behaviour. Physical weathering has an impact on the grain size of the soil. Meanwhile, chemical weathering affects the clay mineral content. The soil swelling behaviour is influenced by the contents and types of clay minerals in it (e.g., [1]). The ability of these clay minerals to absorb water determines the swelling of the soil volume to occur. Nevertheless, the classification of the expansion degree of expansive soils is based on primarily their physical properties.
In Indonesia, the identification of expansive soils should lead to the determination of the soil expansion degree. In the Construction and Building Guidelines [2] and the Earthworks for Roads [3], the classification of the degree of expansion is carried out using several methods. In these two references, the parameters of expansive soils, such as the potential and swelling pressure, as well as degree of expansion can be determined using primarily the soil index properties, through the methods proposed by Skempton (1953), Holtz and Gibbs (1959), Seed et al. (1962), Van der Merwe (1964), and Chen (1965), Chen and Raman (1967) [2,3].
Research on expansive soils in several regions of Indonesia has been carried out. This paper is to review the results of previous research on expansive soils from several locations in Indonesia, and to subsequently evaluate some of the criteria above using the compiled laboratory test results.

Degree of expansion classification
Based on [2,3], the degree of expansion can be determined using soil physical index properties, such as liquid limit (LL), plasticity index (PI), activity (A), and clay contents. The three most widely used methods are the methods proposed by Seed et al. (1962), Van der Merwe (1964), and Chen (1965).

Activity method
The activity method was proposed by Seed, Woodward, and Lundgren based on research in which 23 artificial mixtures of Bentonite, Illite, Kaolinite, and fine sand were used. The samples were compacted at maximum dry density and optimum moisture content with a load of 1 psi [4], and the expansion was subsequently measured during the soaking time. The activity of the object is defined as: (1) In equation 1, C is the percentage of clay finer than 0,002 mm and PI is the plasticity index, the difference between liquid limit (LL) and plastic limit (PL). This method appeared to be an attempt to improve the United States Bureau of Reclamation (USBR), where the shrinkage limit was not included in the criteria for evaluating the swelling potential of soil [4]. The degree of expansion can then be determined using the graph of the relationship between percent clay content and activity.

Van der Merwe's criteria
This method was developed based on research using soils from South Africa [5]. This criterion is one of the most popular criteria used to determine the degree of expansion for expansive soils. A simple classification in determining the degree of expansion can be determined using the percentage of clay content (C), plasticity index (PI), and the activity of the sample. Different from the activity method developed by Seed et al. (1962), in the Van der Merwe's criteria, the activity is determined using the equation proposed by Skempton (1953), which is the ratio between PI and C [6]. (2) The classification of the degree of expansion is limited by the activity, which is divided into low, medium, high, and very high swelling potential.

Chen's criteria
Approaches of using the Atterberg limits to predict the degree and potential of swelling are very common. One of them is Chen (1965) whose developed a correlation to predict the degree of expansion from undisturbed soil. Determination of the degree of expansion can be determined based on percent finer than No. 200 sieve size and liquid limit (LL) [7], and the degree of expansion is defined from low to very high.

Methodology
In this paper, various datasets from seven references from several regions in Indonesia were used: four studies from Central Java, and other studies from West Java, East Kalimantan, and East Nusa Tenggara. This review used data obtained from studies that reported the results of the Atterberg limits and grain size testing. A total of 37 datasets were used to determine the degree of expansion using the activity method and Van der Merwe criteria. Meanwhile, 34 datasets were used to determine the degree of expansion using the Chen criteria. The degree of expansion was determined by plotting the data on each reference graph, and each criterion was then evaluated using the data collected.

Results and discussion
The data collected are plotted against the reference graphs in Figure 1. For the activity method, the data are distributed from low to very high expansion degree, while for the Van der Merwe criteria, the data are distributed from low to very high expansion degree. For the Chen criteria, the data are distributed from medium-high to very high expansion degree. These suggest that a different set of criteria may provide different conclusions for the same soil samples.
It is of interest that Figure 1(c) shows something unique, where the test results for expansive soil from East Nusa Tenggara are outside the dominant distribution of samples. Though not explored in detail in the original reference, this may be due to geological factors, where the parent rocks of the expansive soils in the eastern regions of Indonesia may have been different from those in the western regions.
The degree of expansion for each criterion is then presented in Table 1 for a further analysis. Tests on mineral content were only carried out by two researchers and showed the same results, where montmorillonite became the dominant clay mineral in the test material. The results of the classification using the three criteria indicate further the discrepancy in the degree of expansion generated for each sample. Based on the mapping of the classification results, the three criteria would generate the same expansion degree for only 19% of the samples, and the three criteria would generate completely different expansion degrees for 22% of the samples.
Two criteria would generate the same expansion degree for 59% of the samples. The activity method and Van der Merwe criteria would generate the same degree for 38% of the samples; this percentage is the highest, possibly due to the similarity of properties required. The activity method and the Chen criteria would generate the same degree for only 8% of the samples, while the Van der Merwe and Chen criteria would generate the same degree for only 13% of the samples.  The difference in degree of expansion resulting from each criterion is determined by the initial conditions of the samples in the study when the criteria were made [15]. The activity method developed the results of testing on compacted mixed materials, while the Van der Merwe criteria were developed based on the results of research on natural soils in South Africa, while Chen's criteria are for rocky mountain soils tested at natural moisture content conditions [7]. Based on this background, even using the same indicators does not necessarily result in the same degree of expansion. In addition, the difference may be due to no single standard on determining the degree of expansion [7]. Any similarity in the degree of expansion resulting from the three criteria could be pure coincidence.
When these three criteria are used to predict the degree of expansion on expansive soils in Indonesia, the results would certainly be different and do not necessarily indicate the actual conditions for the reasons discussed above. Indonesia has types of soils produced by weathering volcanic material that do not move geologically called tropical residual soils [16].
Expansive soils are fine grained soil where the clay minerals content affects overall behaviour [17,18]. The clay mineral contents of the soils are influenced by geological and environmental factors where the soils were formed, then determine the characteristics of the expansive soils. The potential and degree of expansion are difficult to predict using general geotechnical parameters because the presence of clay particles would affect the physical and chemical properties [19]. Atterberg limits and clay contents are combined into a parameter called activity [6], however, the activity classes proposed by Skempton do not directly include mineral contents, so they could not predict accurately the swelling potential of the soils.
Based on a simple analysis, it shows that the approach using physical properties alone is not sufficient to determine the degree of expansion that occurs in expansive soils. Several approaches are needed to accurately predict the degree of expansion on the expansive soil. The conditions of the soil samples may have to be explicitly considered when developing such expansion degree criteria.

Conclusion
Based on a review on expansive soils from several locations in Indonesia, there is a discrepancy in the prediction of the degree of expansion between one criterion and another. It seems that physical index properties are not sufficient to predict the degree of expansion that occurs, and the existing criteria cannot accommodate the condition of the soil samples. A more robust classification for the expansion degree of expansive soils is warranted.