1 Introduction

Problematic soils are less reliable for construction as their behavior can change dramatically with change in climatic conditions. Expansive, liquefiable, collapsible and sensitive soils are all under the category of problematic soils (Ali 2011; Rezaei et al. 2012). Collapsible soils are unsaturated and experience a drastic reduction in volume with increase in moisture content (with or without increasing in stress) (Rogers 1995; Houston et al. 2001; Gaaver 2012). Collapsible soils have high strength with resistance to deformation at their natural moisture content. However, with an increase in moisture content, the interparticle forces decrease which result in a reduction in sudden volume (Barden et al. 1973; Houston et al. 1988). Environmental conditions in arid and desert regions are suitable to create collapsible soils, but these soils can also be found in other areas (Houston et al. 2001).

Collapsible soils have a loose granular structure including a small amount of fine particles such as clay. The shape of soil particles is a factor that affects soil particles arrangement and their collapse potential. Spherical particles can possess maximum density condition but due to presence of fines, capillary forces or cementation bonds, the density decrease. However, the particles with angular shape usually create unstable structures with higher void ratios (Rogers 1995).

Collapsible soils have similar physical characteristics such as (Rogers 1995; El Howayek et al. 2011; Ziaie Moayed and Kamalzare 2015):

  1. 1.

    Open physical structure

  2. 2.

    High porosity (more than 40%)

  3. 3.

    Low dry density

  4. 4.

    Low saturation degree (less than 60%)

  5. 5.

    Young geological formations that had less time to obtain their natural density

  6. 6.

    Strongly affected by changes in moisture content or effective stress level

  7. 7.

    High silt content (more than 30%, sometimes up to 90%) and low clay amount

  8. 8.

    Low resistance in intergranular forces

Collapsible soils contain aeolian, alluvial, colluvial and residual deposits created by natural sedimentation under certain environmental conditions. These soils are also created in artificial embankments which are compacted at the dry side of the optimum moisture (Rogers 1995). Loess soils are wind deposited collapsible soils that cover about 10% of Earth’s dry areas (Nouaouria et al. 2008; Gaaver 2012; Li et al. 2016). Researchers reported that loess soils are distributed in places such as America, Australia, Brazil, Europe, Central Asia, China and Russia (Evstatiev 1988; Rogers et al. 1994; Rogers 1995). Loess soils are composed of silt-sized particles and clay. Carbonates and capillary tension provide links between particles, creating an open structure which also has collapse potential with an increase in moisture, stress or both (Barden et al. 1973). Therefore, loess lands are extremely problematic and responsible for soils settlement (Dobrescu et al. 2017).

Technical literature reports various methods to identify collapse potential in soils. In laboratory methods, the criteria are classified into two categories, qualitative and quantitative. The criteria stated by Holtz and Hilf (1961), Feda (1966) and Handy (1973) are qualitative and presented in Table 1 (Feda 1966; Handy 1973; Rogers et al. 1994; Nouaouria et al. 2008). The criteria submitted by Abelev (1948), Jennings and Knight (1975) and Sabbagh (1982) are quantitative and presented in Table 2 (Lutenegger and Hallberg 1988; Rogers et al. 1994; Ali 2011).

Table 1 Diagnosis of collapse attributes using soil engineering properties (qualitative criteria) (Feda 1966; Handy 1973; Rogers et al. 1994; Nouaouria et al. 2008)
Full size table
Table 2 Diagnosis of collapse attributes using the simple oedometer test results (quantitative criteria) (Lutenegger and Hallberg 1988; Rogers et al. 1994; Ali 2011)
Full size table

The qualitative criteria use soil bulk density, natural void ratio, porosity, Atterberg limits, natural moisture content or a combination of them. However, these criteria are unable to definitively predict the soil collapse properties. Quantitative methods apply the results of oedometer tests on collapsible soil samples. In this testing method, the soil sample with natural moisture content is first loaded to a certain level and then immersed with water to measure the deformation or collapse (Houston et al. 1988).

Collapsible soils require remediation to decrease their collapse degree and consequential large deformations. The remediation methods are divided into near surface and deep treatments (Ziaie Moayed and Kamalzare 2015; Jefferson et al. 2015). Choice of improvement method depends on several factors (Houston et al. 2001; Jefferson et al. 2015) such as:

  • Source of water which causes collapse

  • Origin of applied stress

  • Time of collapsible soil’s identification (before or after construction)

  • Cost of improvement methods

  • Depth and size of the collapsible area

Various treatment methods are reported for improving collapsible soils but grouting, dynamic compaction, and hydraulic binder stabilization methods are the most common ones. Hydraulic binder stabilization method is applied to shallow depths up to 5 m (near surface treatment) with adding cement, lime and fly ash (Jefferson et al. 2015). Another type of additive used in this method is a by-product of Portland cement production called cement kiln dust (CKD) (Ismail and Belal 2016). Each year thousands of tons of CKD is produced from furnaces of cement factories and discarded as waste. The CKDs are pollutant to the environment and occupy space in landfill sites. Therefore, a new application of this product can be beneficial from both environmental and economic aspects.

A wide range of soils can be stabilized by the addition of CKD. Miller and Azad (2000) reported that the addition of CKD to moderately and highly plastic soils decreased the plasticity index (PI). Also, the maximum dry unit weight (MDUW) and optimum moisture content (OMC) decreased and increased, respectively. Ismail and Belal (2016) studied the effect of CKD on properties of a young alluvial soil located near Nile Delta in Egypt. The addition of CKD up to 20% decreased MDUW and increased OMC. Al-Refeai and Al-Karni (1999), Albusoda and Salem (2012), Noorzad and Pakniat (2016) all reported that the collapse potential decreases with the addition of CKD.

The purpose of this study was to investigate the geotechnical properties of loess soil in the area of Mohammad-Abad Semnan railway, determine the collapse potential of this soil and investigate the effect of CKD on mitigating collapsibility with regards to the soil type and chemical contents. In this study, the collapse potential of two remolded soil types collected from Mohammad-Abad region was measured using oedometer tests. The soil samples were prepared with similar initial dry density and moisture content. The collapse properties were evaluated with immersion of samples at vertical stresses of 100, 200, 400, 800 and 1600 kPa. Rainfall and surface running water are the main causes of collapse in this region; therefore the use of hydraulic binders should be sufficient to prevent collapse from the surface. Limited research is performed to study the effect of CKD on mitigation of collapsibility, for that reason CKD was used as an additive.

2 Materials and Methods

The collapse potential of two types of remolded soils was studied by preparing a total of 30 soil samples. Soil A was used to prepare 15 samples in the initial moisture content of 20.5% and dry density of 1.61 gr/cm3; and soil B was used to prepare 15 samples in the initial moisture content of 12% and dry density of 1.5 gr/cm3. The samples were placed in the oven (100 °C) to reach the moisture content of 5.7% and 5.4% respectively. Collapse potential of the soil samples were measured at immersion stresses of 100, 200, 400, 800 and 1600 kPa.

The CKD used in this experiment was supplied from the Urmia cement factory. First, Proctor compaction and Atterberg limit tests were performed on the soils mixed with CKD at 4%, 8%, and 12% to investigate the effect of this material on the geotechnical properties of the soils. Then CKD with the percentages of 4%, 8% and 12% (by weight of dry soil) was added to samples prepared from soil A and B and compacted in consolidation ring at their maximum moisture content. The specimens were sealed in plastic sheets to avoid loss of moisture and treated for 7 days. After treatment, the samples were placed in the oven (100 °C) to reach the moisture content of 5.7 and 5.4% respectively. The collapse potential of treated samples were measured at the vertical immersion stress of 200 kPa. The tests on CKD treated soil samples under 200 kPa stress were repeated three times.

2.1 Test Materials

2.1.1 Soil

The soil in the area around Mohammad-Abad located in East of Semnan city which has a hot and dry weather has been split and shown subsidence. Therefore, two soil types in the range of two gaps with 1 km distance where plotted and soil samples were collected from the depth of 0.5–1 m from each location. The first soil sample (soil A) was adjacent to a split with 1.5 m depth and 100 m length and the second soil sample (soil B) was adjacent to a split with 4 m depth and 300 m length. The in-situ soil was dry and there was no possibility to withdraw undisturbed soil samples, therefore remolded soil was used to prepare samples. Geotechnical soil properties such as soil moisture content, grain size distribution, Atterberg limits, specific gravity, maximum dry unit weight, dry bulk density, pH and electrical conductivity were determined using standard laboratory methods (Table 3). Scanning electron microscopy (SEM) was used to study the structure of the soil. Dry bulk density (ρd) was evaluated using paraffin-coated clod method. For this, clods in collected samples were oven dried and their dry masses were measured. The soil samples were immersed in melted paraffin so that coated samples were obtained. The mass of paraffin coated sample was measured and then dropped in a water to measure its total volume. The mass of paraffin and its specific gravity are used to calculate its volume and consequently the volume of soils. Dry mass and volume of the soil are used for calculating its dry bulk density. This test was performed on twelve soil samples, six per soil type, and an average of the value obtained is reported in Table 3.

Table 3 Soil properties
Full size table

2.1.2 Cement Kiln Dust (CKD)

Cement kiln dust is an ash with granular and powder form collected by fabric filters in the exit gasses of Portland cement manufacturing plants. Chemical components of CKD vary with type of furnace, raw material, fuel and collection filter (Adaska and Taubert 2008). The composition of this ash is similar to cement raw materials but the ratio of alkalis, chlorides, sulfates, and other volatile materials in the ash is higher than the raw materials, hence only a small amount of the ash is returned as feed to the cement manufacturing process and plenty of it is excreted (Maslehuddin et al. 2009). The addition of water to CKD results in a behavior similar to Portland cement due to the presence of free lime (Miller and Azad 2000). Therefore, the use of CKD can improve the soil strength at a lower cost in addition to reducing waste management (Miller and Zaman 2000; Ismail and Belal 2016).

Chemical analysis of CKD was performed by x-ray diffraction (XRD) and the results are shown in Table 4. The cementation factor (CaO) is a large proportion of the CKD’s composition.

Table 4 Chemical analysis of CKD used in this experiment by XRD
Full size table

2.2 Test Method

The one-dimensional collapse potential (Ic) of unsaturated soils is estimated using a consolidation apparatus and following ASTM D5333-03. This test is applicable to both disturbed and undisturbed soils. This test is used to measure the magnitude of collapse potential at the occurrence of any particular vertical stress using Eq. (1) according to ASTM D4546-14e1 standard.

$$I_{c} = \left( {\frac{{h_{1} - h_{2} }}{{h_{1} }}} \right) \times 100$$
(1)

where Ic = collapse potential, h1 = soil sample height in immersion stress just before saturation, h2 = final soil sample height in immersion after saturation.

Collapse index (Ie), which is the collapse that happens as a result of immersion under 200 kPa is used to define the collapse degree of soil A and B before and after treatment. Collapse Index is calculated using Eq. (1) and the degree of soil collapse can be classified as shown in Table 5.

Table 5 Classification of the degree of collapse using collapse index in accordance with ASTM D5333-03 standard
Full size table

2.2.1 Remodeled Sample Preparation Method

In this experiment, remolded samples were prepared according to the method described by Haeri et al. (2012):

The soil was screened through Sieve No. 100 and oven dried at 100 °C. The required dry soil for preparing the soil sample with the target dry density, was weighed. Mass of water required for achieving target moisture content was added to the soil. Water and soil were completely mixed and compacted in three layers using a small hammer. The dry density is calculated using the dry mass of soil (Ms) and volume of ring (Vring):

$$\rho_{d} = \frac{{M_{s} }}{{V_{ring} }}$$
(2)

The void ratio is calculated using Gs (given in Table 3) and the dry density:

$$e = \frac{{G_{s} }}{{\rho_{d} }} - 1$$
(3)

Due to the limited densification of soils at their natural moisture content, soils with higher moisture content were compacted in the ring with a small hammer. The prepared samples were first placed in the oven at temperature 100 °C and removed once reaching their natural moisture content.

3 Results and Discussions

Scanning electron microscope (SEM) images taken from soil A and B show that the higher percentage of clay content in soil A created a more integrated and compact structure compared to soil B and therefore a lower porosity (Fig. 1a and b). Presence of larger particles like sand and smaller amount of clay in the texture of soil B created an open structure. The links between grains in the collapsible soils are supported by clay bonds and soil suction. A lower level of clay content results in small contacts between soil grains which can collapse with the presence of water. The SEM image of soil B (Fig. 1b) shows the presence of both angular and round particles. Increase in angularity can increase the formation of a more collapsible structure (Rogers 1995). Fine particles can be observed on the surface of larger particles in both soil types. The larger particles are mainly quartz sand and silt grains but the fine particles can be clay or carbonate. Presence of carbonate is more presumable in soil B due to the lack of clay.

Fig. 1
figure 1

SEM images of two types of soils with a magnification of 50 micrometers, a soil A, b soil B

Full size image

Element analysis is performed on soil A and B to investigate their components and their possible effect on their collapsibility. The results are presented in Table 6. The high atom percentage of the aluminum element in soil A indicates the presence of higher clay content compared to soil B which is in agreement with the grain size distribution presented in Table 3. The low amount of clay in soil B creates open pores in the soil structure. The sodium content in soil A is less than soil B. Sodium ions (salt) create a tendency to absorb water during soaking, therefore, increasing collapse potential in soil B. The larger sized non-hydrated calcium and magnesium divalent cations compared to sodium ion reduce repulsion forces between clay particles. Therefore, with the presence of divalent cations (calcium and magnesium) compared to a monovalent cation (sodium), clay particles can get closer. Soil B contains lower clay content and higher sodium ions; therefore, the effect of calcium and magnesium ions for the purpose of grain accumulation is less compared to soil A. On the other hand, the presence of calcium in soil B (which has a lower level of clay content) can be an indication of calcium carbonate bonds between particles. A similar conclusion was made by Derbyshire et al. (1995) on Malan loess located in China. Their soil had an open structure (large voids) and contained a low amount of clay, but more salts which is similar to soil B. The SEM images of undisturbed Malan loess showed that with the absence of clay matrix, carbonate cementation plays the link between the particles.

Table 6 The existing elements in the soil A and B, analysis by scanning electron microscopy/energy dispersive x-ray spectroscopy (SEM/EDS)
Full size table

3.1 Collapse in Remolded Samples

The results obtained from collapse tests on remolded soil A and B are presented in Fig. 2. The results show high collapsibility in both soil types, although both are remolded.

Fig. 2
figure 2

Collapse potential in remolded soil samples, a soil A, b soil B

Full size image

The degree of soil collapse at different immersion stresses for soil A and B is estimated using the classifications presented in Table 2 by different researchers (Table 7). According to the ASTM D5333-03 standard rating, the degree of collapse in immersion stress of 200 kPa in soil A is moderate and in soil B is moderately severe. The two criteria of Jennings and Knight (1975) and ASTM D5333-03 show to have the best assessment for collapse degree.

Table 7 The degree of collapse in soil samples using the classification presented in Table 2
Full size table

The collapse degree of soil A and B is defined with the geotechnical properties of the soils and applying criteria presented in Table 1, and the results are presented in Table 8. The collapse potential of soils is different compared to when laboratory experiments are performed and applied to define collapsibility. Therefore laboratory test results are the most reliable to evaluate collapse potential.

Table 8 Diagnosis of collapse attributes using geotechnical properties of the soil
Full size table

3.2 CKD Stabilized Soil Properties

The addition of CKD can change the soil properties such as Atterberg limit, maximum dry density and optimum moisture content. The Atterberg limit of the soil A and B were evaluated after adding 4, 8 and 12 percent of CKD to the dry soil. Changes in liquid limit (LL), plastic limit (PL), and plastic index (PI) of the two soil types with a change in CKD are presented in Fig. 3. The results show different trends for change in Atterberg limits with the addition of CKD for soil A compared to soil B.

Fig. 3
figure 3

Changes in the liquid limit (LL), plastic limit (PL) and the plasticity index (PI) of two soil types with an increase in CKD, a soil A, b soil B

Full size image

The LL and PI both decreased with the addition of 8% CKD in soil A and then increased at higher CKD content (Fig. 3a). Decrease of LL may be due to the high amount of calcium oxide (CaO) available in CKD (Al-Refeai and Al-Karni 1999; Albusoda and Salem 2012). Also, increase in LL can be due to the additional water needed to bring the soil and CKD to a liquid state. Researchers such as Al-Refeai and Al-Karni (1999), Parsons and Kneebone (2004) and Albusoda and Salem (2012) also reported the decline of LL with increasing CKD up to 8% in their research. The remarkable point is that collapsible soils in similar regions such as Iran, Iraq and Saudi Arabia have the same results with combining CKD. In soil B, the increase in CKD content resulted in a continuous increase in LL and PI (Fig. 3b). Soil B has lower clay content (Table 3) and a lower PI which results in higher amounts of moisture content to initiate the reaction between CKD and soil particles. Therefore, soil B has a higher level of LL after mixed with CKD compared to soil A. Parsons and Kneebone (2004) studied on several types of clay, silt, and sandy soils and observed that the liquid limit may decrease or increase with the increase in CKD. So, the effect of CKD on the behavior of different soil types may have different results.

Compaction test ws performed on soil A and B after mixed with a different percentage of CKD. The summery of Proctor compaction test results are presented in Table 9. According to Fig. 4a, by adding 4% CKD to soil A, the optimum moisture content (OMC) decreased and maximum dry unit weight of the soil (MDUW) decreased slightly. With the addition of CKD up to 8% and 12%, the OMC gradually increased and MDUW decreased. A similar trend was observed in studies such as Albusoda and Salem (2012). The addition of 4 and 8 percent CKD to soil B (Fig. 4b) resulted in an increase in OMC and reduction in MDUW. By combining 12% CKD, the OMC slightly decreased and the MDUW of the soil slightly increased. The increase in OMC is probably due to the additional water needed to increase the action of cementation between soil and CKD which facilitate the compaction of soils especially for soils with low clay. Many researchers reported a similar behavior of collapsible soils with addition of CKD, where OMC increased and MDUW decreased (Baghdadi et al. 1995; Miller and Azad 2000; Sariosseiri and Muhunthan 2008; Carlson et al. 2011; Albusoda and Salem 2012; Ismail and Belal 2016).

Table 9 The summery of Proctor compaction test results on soil A and B with various CKD contents
Full size table
Fig. 4
figure 4

The results of the compaction test of stabilized soils with different percentages of CKD, a soil A, b soil B

Full size image

3.3 Collapsibility of CKD Stabilized Loess

Collapsibility potential of treated samples with different percentage of CKD (4, 8 and 12%) were measured at a vertical immersion stress of 200 kPa (Fig. 5). The collapse degree of CKD treated samples were identified using collapse index and ASTM D5333-03 classification. The results are presented in Table 10.

Fig. 5
figure 5

The results of the collapse potential of stabilized samples with different percentages of CKD, a soil A, b soil B

Full size image
Table 10 Determining the degree of stabilized samples using the collapse index based on ASTM D5333-03
Full size table

The collapse potential of soil A decreased with the addition of 8% and then increased with the addition of 12% CKD (Fig. 5a). This behavior can be due to increased levels of alkaline materials such as sodium in percentages higher than 8% which causes dispersion in clay particles and an increase of collapsibility. Therefore, the optimum amount of CKD to reduce collapsibility of soil A is 8%. The collapse potential of soil B decreased continuously with the addition of CKD up to 12% (Fig. 5b). According to the grain size distribution presented in Table 3, soil B contains less amount of clay compared to soil A. Therefore, the addition of CKD to collapsible soil with lower clay content increased adhesion between the soil grains. Increase in curing time can also play a role in mitigating the collapsibility to optimize the role of CKD treatment.

In summary, the collapsibility of soil B decreased more pronouncedly compared to soil A with the addition of CKD. The results show that there is a clear correlation between the plasticity index (PI) and change in collapsibility potential with the addition of CKD.

The collapse degree and its changes with the addition of CKD are presented in Table 10 for soil types A and B. The collapse degree of soil A has changed from moderate to slight from 4 to 8% CKD and to moderate with the addition of CKD to 12%. The collapse degree of soil B changed from moderately severe to moderate by adding up to 12% CKD. The collapse index could be reduced up to 52.09% and 57.13% in soil A and B but at different percentages of CKD. Al-Refeai and Al-Karni (1999) improved a collapsible soil (17% clay) located in Saudi Arabia with CKD. By adding 5%, 10% and 15% CKD to two soils with different dry densities, they reported a significant reduction in collapse potential of the two soil types. Also with the addition of more than 10% CKD, the collapse potential remained constant. Albusoda and Salem (2012) stabilized dune sand (less than 10% clay) located in southern Iraq by using CKD. In their study, the addition of CKD up to 12% reduced the collapse potential up to 70%. Also in their study, the low amount of clay was a major factor in the adhesion of soil aggregates when CKD increased.

The effect of CKD on collapse potential depends on the soil type, gradation, clay content, curing time and also chemical contents of CKD. The results show that CKD can be a suitable material for reducing the collapse potential. The amount of sodium in CKD show different results and finally, CKD can be a suitable material to reduce the collapse potential of the loess soils.

4 Conclusions

Collapsible soil samples were collected from Mohammad-Abad railway station which had shown split and subsidence. The samples were collected from two different locations with different split and subsidence values to investigate the reason behind their different collapse behavior. The potential of using cement kiln dust (CKD) on improving their collapsibility investigated. The results from this study are summarized as the following:

  1. 1.

    The clay content in collapsible soils has an impact on collapse potential. A lower level of clay content can increase the collapse potential. Less amount of clay results in a small contact between soil grains which can collapse with the presence of water.

  2. 2.

    The collapse potential of the collected samples at the two locations was evaluated at different immersion stresses using one-dimensional consolidation apparatus. According to ASTM D5333-03 and Jennings and Knight (1975), the soils are collapsible.

  3. 3.

    The Atterberg limits were tested for the two soil types before and after adding CKD. The liquid limit (LL) can increase or decrease with increase in CKD content. In most cases with the addition of CKD to collapsible soils, the optimum soil moisture increased and the maximum dry unit weight of the soil decreased.

  4. 4.

    The optimum moisture content and maximum dry unit weight both changed with adding CKD.

  5. 5.

    The addition of CKD to both soil types could decrease the collapse potential more than 50%. However, the CKD content can be different for this level of reduction for different soil types.

  6. 6.

    The amount of clay in soil and sodium present in the CKD, when stabilized with CKD, has a major impact on the potential collapse of loess soils. When the clay content is low, the impact of sodium ions in the CKD that causes more dispersion in clay particles will be less. As a result, with increasing CKD the potential collapse will significantly reduce in soils with less clay.