Green coal mining is one of the main paths to achieving the "double carbon goal" in China's coal industry, in which the repair of ground fractures in coal mining collapsed areas is an important part of green coal mining; research on this topic has important scientific significance and engineering value (Fan 2019; Li and Ai 2019; Dai et al. 2020; Wang et al. 2021; Cai et al. 2023). Particularly in northern Shaanxi, which has entered a large-scale mining stage in the last decade (Fan et al. 2019; Chen et al.2022; Teng et al. 2023), some problems need to be solved regarding the repair of ground fractures in the typical geomorphological conditions of the region. Specifically, similar technology is used to repair the characteristics of coal mining ground fractures at different geomorphologies, which does not achieve the effect of precise governance. Various types of chemical curing materials, which have a certain impact on water and soil, do not achieve the green governance effect; thus, governance remains on the shallow surface. The treatment is limited to shallow surface fractures, and the treatment of hidden fractures is relatively limited and has not achieved the full treatment effect. Therefore, based on the fracture characteristics in multi-faceted landforms, studying green microbial-induced calcite precipitation (MICP) technology to treat coal mining fractures has some practical value (Hou et al. 2019; Choi et al. 2020; Ju et al. 2020; Li et al. 2021).

There is a wealth of research on the repair of fractures at coal mining sites, which can be generally divided into the following three areas. Firstly, the self-healing of coal mining ground fractures is currently considered to have a two-fold mechanism. On the one hand, a comprehensive study by several disciplines concluded that as the coal mining face advances, some of the coal mining ground fractures would close under the additional stress of mining (Li et al. 2017a; Zhang et al. 2017; Xu 2020;), and the span of the coal mining fracture will become smaller after closure when the effect on soil moisture content around the fracture is limited (Liu et al. 2019; Bai et al. 2020), which is considered to have achieved the self-healing effect. On the other hand, some scholars found that when the ground surface is loess, it can swell to repair the ground fractures after water absorption due to its richness in expansive clay minerals (Li et al. 2017b; Liu 2018; Huang et al. 2019).

Secondly, for ground fractures in the exposed ground, the current mainstream treatment measure in coal mines is to use mechanical equipment to fill the topsoil with earthworks. This type of repair technology can also be divided into two areas. On the one hand, spot repairs are performed in conjunction with topographical features for zones with obvious fractures, and the Wu et al. (2022) team has achieved outstanding results in this regard. Conversely, for areas with gentler terrain, which are suitable for unified restoration by large-scale mechanical equipment, the Hu and Yuan (2021) team has added humus, river mud, and other beneficial substances in appropriate amounts to achieve rapid recovery of the restored vegetation.

Thirdly, for coal mining fracture curing restoration, research on this topic can be broadly divided into two aspects: chemical material curing restoration and plant curing restoration (Song et al. 2022). On the one hand, some scholars have developed ultra-high water curing materials, which are favored due to their strong operability (Ren and Yu 2022). In contrast, based on microbial technology, the Bi et al. (2021) team has achieved efficient vegetation restoration in coal mining collapsed areas.

In summary, coal mining land fracture restoration techniques have shown diversified development and have achieved better results under different conditions. However, accurate, green, and efficient restoration under multiple geomorphological conditions still faces challenges. MICP technology is a new green restoration technology that has been applied on a large scale in the field of geotechnical restoration in recent years (Leng et al.2019; Pitcha et al. 2019; Zheng et al. 2019; Liu et al. 2020; Zhang et al. 2022) and deserves further study.

Based on the above-mentioned concerns, this study used a coal mine with multiple geomorphologies as an example, and we applied MICP technology based on the geometry, filling, and stress characteristics of coal mining ground fractures. A combination of macro-mechanical, hydrological, and souring experiments was performed on an indoor experimental study, and we also analyzed the effect of MICP repair on ground fracture soil samples from different landforms (the overall technical flowchart is shown in Fig. 1).

Technical flowchart

2.1 Overview of the study area

The study area was the Lime Strip Tower Mine, which is located in the southern part of the northern Shenmu mining area of the Jurassic coalfield in northern Shaanxi Province and the central part of Shenmu County, Yulin City, Shaanxi Province, China. The Yellow River, Kaokaowusu Gully (a grade 2 tributary and the lowest point in the study area, elevation 942.2 m), flows in the middle of the mine. To the north of Kaokaowusu Gully is a blown-sand region landform, and to the south is a Loess ridges landform, with rivers scattered across both types of landforms. The river is located in a valley landform.

The first coal seam in the study area varies slightly from north to south, with some areas in the north first mining 1^–2 of coal at a thickness of about 2.0 m and other areas first mining 2^–2 of coal at a thickness of about 5.0 m. Areas in the south first mined 2^–2 of coal at a thickness of about 4.5 m.

The coal seams are all developed in the Yan'an Formation, which is stratified from bottom to top by the sandstone Zhiluo Formation, the mudstone Anding Formation (localized in the north), the laterite Baode Formation, the loess Lishi Formation, the Salausu Formation, and the Quaternary wind-deposited sand (alluvium in the valley). The Yan'an, Zhiluo, and Anding Formations form a bedrock thickness of approximately 40 m above 1^–2 of coal and 60 m above 2^–2 of coal. The laterite Baode Formation and the loess Lishi Formation combine to form the impermeable layer, which is approximately 70–110 m thick. The Salausu Formation and the Quaternary wind-deposited sand form the sand layer in the northern part of the study area, with a total thickness of about 20 m. The above strata are nearly horizontal in the study area, with no tectonic development in the area, and the types of rock masses overlying the coal seam are essentially the same. At present, coal mines have entered the stage of large-scale mining. There is a large number of ground surface fractures in the north and south. Different landforms show differences in coal mining ground fractures, and targeted repair work is required.

2.2 Characterization of fractures in coal mining sites in MICP-oriented multifaceted landforms

The geometry, filling, and stress characteristics of fractures in coal mining sites were investigated to provide key parameters for the production of fracture soil samples for indoor experiments.

2.2.1 Geometric and filling characteristics of fractures in coal mining sites in multiple landforms

As previously mentioned, the landforms in the study area included three landform types, namely the blown-sand region landform, the loess ridges landform, and the valley landform. The coal mining ground fractures in these three landform types were studied, mainly including geometric features and filling characteristics. Since many factors can affect the developmental characteristics of coal mining fractures, the topography and other influencing factors were not considered here (all parameters were chosen to be counted in flat terrain. According to Hou et al. (2021), the influence of the terrain on the geometric parameters of coal mining ground fractures only accounts for 16% of the total number of ground fractures) in order to rationalize the indoor experimental model. Ground fractures in the area could be divided into two categories: boundary fractures and internal parallel fractures, with the number of boundary fractures accounting for 14.3%–26.3% and the number of parallel fractures accounting for 72.7%–85.7%. Based on these percentages, the geometric and filling characteristics of coal mining ground fractures were investigated during the coal mining advancement and ceasing periods.

2.2.2 Geometrical characteristics

The geometrical characteristics of the coal mining fractures mainly included the depth and width of the fractures, while the fracture length, spacing, and drop were not included in the scope of this study in order to construct elements for the MICP experimental samples.

2.2.2.1 Depth

1. (a)

   Parallel ground fractures (parallel to the open-off cut)

Field observations were performed on four groups of parallel ground fractures, including Group 1 for coal seams 2^–2 mined in the blown-sand region landform, Group 2 for coal seams 2^–2 mined in the loess ridges landform, Group 3 for coal seams 1^–2 mined in the loess ridges landform and Group 4 for coal seams 1^–2 mined in the valley landform. The results of depth variation of parallel ground fractures using field geological measurements (time point 0 was 20 m away from the overrun fracture, time point 2 d was 10 m away from the overrun fracture, and time point 4 d was close to the fracture) are shown in Fig. 2. It is apparent that under the same coal seam mining thickness conditions, the depth of parallel fracture development in the loess landform at the time of stabilization was 1.93 times that in the blown-sand region landform, and the depth of parallel fracture development in the valley landform was 1.50 times that in the loess landform.

Dynamic variation diagram of ground fracture depth

1. (b)

   Boundary fractures (parallel to the trench)

Field observations were made on four sets of boundary fractures, of which 5–8 were observed in the same environment as the parallel fractures. The results of the depth variation of the boundary ground fractures by field geological measurements (time point 0 was 20 m away from the overrun fracture, 2 d was 10 m away from the overrun fracture, and 4 d was close to the fracture) are shown in Fig. 2. It can be seen that under the same coal seam mining thickness conditions, the depth of boundary fracture development in the loess landform at the time of stabilization was 2.06 times that in the blown-sand region landform, and the depth of boundary fracture development in the valley landform was 0.61 times that in the loess landform.

1. (c)

   Depth development pattern

It is evident from Fig. 2 that the boundary fractures developed to a greater depth than parallel fractures under the same conditions. The overtopping distance of the 2^–2 coal ground fracture, which is thicker, was greater than 20 m, while the overtopping distance of the 1^–2 coal ground fracture was about 10 m. The overall greater depth of the loess landform compared with the blown-sand region landform under different landform conditions is in accordance with the Moore-Cullen criterion for the ultimate depth of development of the soil critical surface. Based on Eq. (1), the ultimate depth of ground fractures was calculated to be 2.1 m for the Salausu Formation and 14.3 m for the loess land fractures. Boundary fractures exceeding the ultimate depth of development will collapse, causing the fracture to disappear(e.g., Fig. 3, demonstrates the eolian study area and the limit of the soil critical surface). The ultimate depth of ground fractures was calculated as follows:

where \({h}_\text{j}\) is the ultimate depth of the ground fracture in the formation in m; \(c\) is the cohesion of the formation in MPa; \(\varphi\) is the angle of internal friction of the formation in °; γ is the capacitance of the formation in kN/m³.

State of the stratigraphic limit prograde of typical landforms in the study area

2.2.2.2 Width

Width of fractures in coal mining ground using field geologic measurements, and the results are shown in Fig. 4. Our results showed that the width and depth of ground fracture development were positively correlated with the stage of fracture development. However, the correlation was low for the recovery period, as it was difficult to connect the depth of fracture development in a short time period, while the opposite was observed for parallel fracture development and was even close to disappearing, with a maximum reduction of 83.3% (connection of parallel fractures in the valley landform).

Dynamic variation of ground fracture width

2.2.3 Fill characteristics

For the MICP experimental sample setup, the filling characteristics of the coal mining ground fractures mainly included fill type, fill density, and pH of the fill.

2.2.3.1 Type and density of the fill

As the repair of coal mining ground fractures occurs after coal mining collapse has been stabilized, this study transpired more than 1 year after the coal mining workings had been completed. The results of the survey are shown in Table 1. It can be seen that parallel fractures had no filler due to limited coal mining openings. The boundary fractures had basically filled material due to the large coal mining openings, with the highest degree of filling due to the fast deposition rate of windy sand under the blown-sand region landform, reaching 100%, followed by the valley landform (22.8%), while there was minimal filling in the loess ridges (9.4%), most of which was still produced by the collapse of loess fractures. The density of all the fracture fills was slightly lower relative to the natural density.

2.2.3.2 pH of the filling

The pH of the mixture was determined by mixing the different fillings with the microbially cured restoration solution and setting the fill weight ratios of wind-deposited sand:loess at 1:0, 0.5:1, 1:1, 1.5:1, and 2:1. The fill weight ratios of eolian sand:alluvium was 1:0, 0.5:1, 1:1, 1.5:1, and 2:1. The results are shown in Fig. 5, and the pH ranged from 7.9 to 10.4, which is suitable for restoration by microorganisms.

pH of the filling and restoration fluid mixture

2.2.4 Stress state characteristics of coal mining shallow surface fractures

The stress characteristics in the shallow surface loose layer of the coal mining fractures could be divided into four zones (Fig. 6a): zones 1 and 3 were stress compression zones, and zones 2 and 4 were stress tension zones. Shallow surface ground fractures were developed within zones 1 and 2, zone 3 had not yet developed any ground fractures, and zone 4 was within a certain depth range and did not belong to the category of ground fractures. Therefore, the stress state of the coal mining fractures could be divided into two types, i.e., boundary fractures in a unidirectional tensile stress state and parallel fractures in a unidirectional horizontal pressure stress state.

a Stress state of shallow coal mining b The water pressure test

In order to identify the stress state of the coal mining ground fractures in the study area, water pressure tests were performed in a borehole in zone 1 (Fig. 6a). A 3–10 m depth was used for the water pressure test with the loess Lishi Formation. The results of the water pressure test are shown in Fig. 6b. It is evident that the maximum stress under the superposition of natural stress and additional stress field was between 0.4 and 0.6 MPa.

3.1 Sample preparation

Based on the results of the coal mining ground fracture study mentioned in Sect. 2.2, the loess fracture remodeling samples were prepared considering the type of ground fracture (parallel vs. boundary ground fracture) and the type of landform (blown-sand region, loess ridges, and valley). Because different landforms are included in the loess layers, for comparative analysis, the fracture in the loess layer was selected as the research object, and the geometry, filling, and stress characteristics of the fracture in different landforms were used as variables. The fracture geometry parameters, filling, and stress design parameters of the samples are shown in Table 2. Based on the findings in Sect. 2, the depth:parallel loess fractured landforms were considered to be twice as deep as the blown-sand region, the valley features were considered to be 1.5 times as deep as the blown-sand region, the loess boundary fractured landform were considered to be twice as deep as the blown-sand region, and the valley features were considered to be 0.6 times as deep as the blown-sand region. The width was positively correlated with depth, and we used 0.5 times the depth for calculations. For each type of sample three parallel samples were constructed for each test item. The fracture models of the samples were created using 3D printing technology. Thin slices of the fracture model were pressed into the soil samples to create fracture types and filled according to the filling parameters in Table 2. Finally, the filled fracture soil samples were injected with a microbial solution (Bacillus megaterium liquid with an OD600 value > 2.0) and cementing solution (0.1 mol/L each of urea and calcium source) at a 1:1 ratio and repeated three times. After 7 days of maintenance (Fig. 7), the next tests of mechanics, permeability and scouring.

Flow of fractured soil sample preparation

3.2 Experimental methods

The objective of fracture repair at coal mining sites is to improve the mechanical and hydraulic properties of the fractured soil, thereby reducing the incidence of soil erosion and geological hazards. Therefore, in this experiment, the parallel fracture repair samples were subjected to unconfined compressive strength tests and variable head infiltration tests, and the boundary fracture repair samples were subjected to triaxial confined compression tests and triaxial seepage tests.

In addition, in order to test the effect of reduced soil erosion in the fracture soil samples after MICP repair, souring experiments were performed using the experimental setup illustrated in Fig. 8 (Leng et al. 2019) as follows: firstly, a loess slope (slope angle 5°) was prepared. Secondly, two fractures were excavated in this slope. Thirdly,the fractures were filled (filling material as a variable), and MICP repair was performed (repair parameters were the same as Sect. 3.1). Finally, a 1 h souring experiment was performed to simulate rainfall (10 mm/h) in the study area (Fig. 9).

Experimental setup for the souring experiment

Illustration of the souring process

4.1 MICP mechanical experiments

4.1.1 Compressive strength of the restorations under different geomorphological features

Based on the relevant parameters in Table 2, the proportion of filling materials was specifically selected for each landform type, i.e., the filling materials for the blown-sand landform were aeolian sand, the filling materials of the loess ridges landform were wind-deposited sand and loess with a mass ratio of 1:1, and the filling materials of the valley landform were wind-deposited sand and alluvial with a mass ratio of 1:1. The results of the compressive strength tests of the restorations under different geomorphological features are shown in Fig. 10.

Mechanical strength of MICP restoration under different geomorphic characteristics

The MICP compressive strength results showed that due to the presence of the remolded fracture-free soil sample pressure, the repair effect of parallel fractures was generally better than that of boundary fractures in terms of mechanical strength; the minimum strength of the repaired soil sample could reach the strength of the remolded fracture-free soil sample, indicating that all landform types could meet the repair requirements in terms of mechanical properties. However, there was a difference in the repair effect of soil strength under the unconfined and confined conditions: the strength of the soil under confined conditions (for boundary fractures) was negatively correlated with repair, and the strength of the soil under unconfined conditions (for boundary fractures) was negatively correlated with the depth of the repaired fracture, suggesting that the stress concentration effect was greater than the repair effect when the compression caused tensile damage. In contrast, the strength of the soil under confined conditions (for parallel fractures) was positively correlated with the depth of the repaired fracture, showing that the repair effect was greater than the stress concentration effect when the compression caused shear damage.

Based on the theory of the interface shear effective coefficient between the MICP-filled repair and the soil, the interface shear effective coefficient includes two types of cohesion and friction effective coefficients, as calculated with Eqs. (2) and (3):

where E_c is the effective coefficient of cohesive force on the interface; E is the effective coefficient of the friction; C₁ is the internal cohesion of the MICP-reinforced filled soil; C₂ is the internal cohesion of the remolded soil; δ is the internal friction angle of the MICP-reinforced filled soil; and φ is the internal friction angle of the remolded soil.

The ratios of the cohesion and internal friction angle of the MICP-filled soil and the solid of the loess body were tested and showed to be > 1.0 (Table 3), indicating that the restoration effect was greater than the stress concentration effect under the conditions of this restoration method. However, for a ratio close to the critical value like the valley geomorphology, another situation may exist for in situ restoration (i.e., high local parameters for the in situ soil, such as internal cohesion and internal friction angle), which requires multiple reinforcements of restoration to further enhance the shear strength parameters.

4.1.2 Compressive strength of different filling materials under the same geomorphic features

The research results in Sect. 2.2 showed that there were no clear natural fillings in the loess ridges and valley landforms; thus, different filling ratios could be selected for the restoration, with the following four ratios tested in this study: eolian sand:loess (alluvium) ratios at 0.5:1, 1:1, 1.5:1, and 2:1. The confined/unconfined compressive strength test results of different filling ratios under the same landform characteristics are shown in Fig. 11.

Mechanical strength of MICP restorations under different filling conditions

The experimental results showed that overall, as the soil filling ratio increased, there was a trend of increasing strength followed by a decrease in strength, which was close to the pH in Fig. 5, indicating that the filling strength is highest at a ratio of 1:1 to 1.5:1. Current studies believe that microorganisms are more active at this pH range and there is a clear increase in calcium carbonate yield (Sun and Miao 2020).

4.2 Hydrological experiments

4.2.1 Permeability coefficients of the restorations under different geomorphological features

Next, we determined the permeability coefficients of the MICP restored confined/unconfined compressive strength tests under different geomorphological features, and the results are shown in Fig. 12.

Permeability coefficients of MICP restorations under different topographic features

The experimental results show that due to the presence of confining pressure, the repair effect of parallel fractures was generally better than boundary fractures in terms of the permeability coefficient. The maximum permeability coefficient of the repaired soil sample was close to that of the remolded fracture-free soil sample, indicating that all landform types can meet the repair requirements in terms of hydraulic properties. The effect of the permeability coefficient of the soil under confined/unconfined conditions was overall similar, i.e., the permeability coefficient of the restorations decreased when the depth of the repaired fractures was deeper and broader.

4.2.2 Permeability coefficients of different filling materials under the same geomorphological features

Based on the aforementioned results (Sects. 2.2, 4.1, and 4.2), this study tested four blown-sand regions:loess (alluvium) ratios of 0.5:1, 1:1, 1.5:1, and 2:1. The results of the permeability coefficient for the confined/unconfined tests with different fillings and the same geomorphological features are shown in Fig. 13.

Permeability coefficient of MICP restoration under different filling conditions

The results showed that the influence of the filling characteristics was greatest on the restoration permeability, especially for the sand-to-soil ratio of 2:1, where the permeability coefficient increased significantly, while the sand-to-soil ratio of 1:1 and below was negligible, and the measured results were close to the permeability coefficient of the loess.

4.3 Souring experiments

Based on the relevant parameters in Table 2, the proportion of filling materials was specifically selected for each landform type, i.e., the filling materials of the blown-sand region landform were wind-deposited sand, the filling materials of the loess ridges landform were wind-deposited sand and loess with a mass ratio of 1:1, and the filling materials of the valley landform were wind-deposited sand and alluvial with a mass ratio of 1:1. The results of the precipitation test of the MICP-restored soil under different geomorphological features are shown in Fig. 14.

Scour volume of MICP-restored soil under different geomorphic characteristics

The experimental results showed that the simulated precipitation of the fracture restoration slope was closely related to the geomorphological features, i.e., precipitation for the blown-sand region > loess ridges > valley under the same topographic conditions, with a relatively similar precipitation amount for the loess ridges and valley.

According to the research results on the scouring mechanism of clay soils at home and abroad (Deng 2021), the scour volume is determined by the interaction between the scour flow capacity and the resistance of the soil, i.e., two levels of external and internal factors. Since the slope and rainfall of this experiment were similar, the flow rates and velocities were also similar, i.e., uniform exogenous factors. Therefore, the simulated scour volume was mainly related to endogenous factors such as the shear strength and pore ratio of the MICP-restored soil. Combined with the test results in Sect. 3.1, both the shear strength and the pore ratio of the blown-sand region restorations were the largest, where an increase in shear strength effectively resisted scouring and a larger pore ratio made the soil looser and more susceptible to scouring. It is evident that the pore ratio of the restored soil was the key factor influencing scour volume in the different landforms, which was closely related to the choice of filling material and the degree of MICP restoration.

However, the exogenous factors were not the same under true conditions, i.e., the topographic slope of the loess ridges was generally greater than that of the blown-sand region (according to Bernoulli's equation, the scour flow velocity would be substantially higher), and the scour flow velocity of the valley was also greater than that of the blown-sand region; thus, it is likely that the blown-sand region scours the least under actual conditions.

The experimental study revealed important findings regarding fracture characteristics and MICP repair in coal mining sites within diverse landforms. Based on our results, the following conclusions were made:

1. (1)

   The depth of parallel fracture development was 1.93 times greater in the loess landform compared with the blown-sand region landform under identical mining conditions. In contrast, the valley landform had 1.50 times deeper parallel fracture development than the loess landform. Additionally, the depth of fracture development at the boundary of the loess landform was 2.06 times greater than that in the blown-sand region, while the valley landform exhibited 0.61 times greater depth than the loess landform. A positive correlation between fracture width and depth was observed across different landforms.

2. (2)

   Confined coal mining openings resulted in no significant filling of parallel fractures. Boundary fractures had the highest filling levels, reaching 100%, due to larger coal mining openings and the fast deposition rate of wind-blown sand in the blown-sand region landform. The valley landform exhibited moderate filling, whereas the loess ridges area had a minimal filling of 9.4%. The density of all fracture fillings was slightly lower than the natural density. The pH of the fillings ranged from 7.9 to 10.4, creating a weakly alkaline environment suitable for microbial restoration.

3. (3)

   Coal mining fractures experienced two types of stress states: boundary fractures that underwent unidirectional tensile stress and parallel fractures that were subjected to unidirectional horizontal pressure. The water pressure tests determined that the confining pressure of parallel fractures was 0.4–0.6 MPa.

4. (4)

   The minimum strength of the repaired soil sample reached the level of the remolded fracture-free soil, indicating that all landforms met the mechanical requirements for repair. However, localized stress concentration effects could surpass the restoration effect during in situ restoration, particularly for the valley landform. Additionally, an optimal filling strength was observed for the sand-to-soil ratio of 1:1 to 1.5:1.

5. (5)

   The maximum permeability coefficient of the restored soil sample closely matched that of remolded fracture-free soil, indicating that all landforms met the hydraulic requirements for restoration. Filling characteristics had the greatest influence on restoration permeability, with a significant increase for the sand-to-soil ratio of 2:1, while ratios of 1:1 and below exhibited relatively stable permeability coefficients.

6. (6)

   Simulated precipitation of fracture restoration slopes was influenced by the geophysical features. Under similar topographic and rainfall conditions, the blown-sand region exhibited the highest precipitation volume, followed by loess ridges and the valley landform. However, practical conditions such as soil scour external factors (scour flow rate and flow) may lead to situations where scour volume in the blown-sand region landform is not minimal.