Sage Journals: Discover world-class research

Abstract

In order to improve the accuracy of the surface dynamic prediction model in mining areas with thick unconsolidated layers and improve Knothe time function, the influence coefficient was firstly changed into the coefficient in exponential form, and the influence coefficient of unconsolidated layer was added. Then, a subsidence basin prediction model for mining under thick unconsolidated layers was established. Next, the model was combined with the improved Knothe function, thus constructing a new mining subsidence prediction model. The new subsidence prediction model was applied in 1414 (1) working face in Huainan mining area. The results showed that the integrated model could better reflect the subsidence process, and the prediction values and the measured values agreed well.

Keywords

Mining subsidence Boltzmann function Knothe time function thick unconsolidatd layers parameters

Introduction

In recent years, it has been found out that the surface subsidence in some mining areas with thick unconsolidated layer in China, such as Huainan, Huaibei, Datun, Yanzhou and etc, is intense and concentrated. Moreover, the convergence of surface subsidence basin edge is slow. Under such background, it is of great significance to dynamically get surface movement laws in such areas. This could reveal the mechanism of mining subsidence, but also provide data basis for buildings (structures) protection and damage identification.

At present, dynamic prediction methods of surface subsidence mainly include time function model, prediction model based on measured data, mechanical and numerical simulation method and etc. Zhang et al. (2020) proposed an improved two-parameter time function model and a simple parameter calculation method; Wang et al. (2012) proposed a time function model of surface subsidence based on Morgan-Mercer-Flodin (MMF) function, and verified its feasibility by engineering application; Djamaluddin et al. (2011) proposed a dynamic prediction model by integrating GIS system and Knothe function and applied it to the prediction of surface subsidence disasters and the evaluation of damage levels; Unlu et al. (2013) used geographic information technology to simulate the dynamic process of surface movement and deformation; Wang et al. (2018) provided a surface prediction model for rock grouting and filling mining and achieved good prediction results; Yang et al. (2017) proposed to use shape variables obtained by InSAR to calculate parameters of the logistic model in view of low quality of SAR image data under certain conditions, so as to conduct prediction by the logistic time function model; Zhao (2005) deduced the functional relationship between dynamic subsidence and time by using the principle of rheological mechanics, and estimated mechanical parameters based on the measured data; Based on the Knothe time function and Zero model, Yu et al. (2003) established a rheological model of overburden movement and deformation; Li et al. (2019) considered the influence of mechanical parameters of numerical simulation software during mining on the prediction results, and constructed a new method of predicting surface settlement. Many scholars have also developed other dynamic prediction methods of surface subsidence (Kowalski, 2001; Nie et al., 2015; Rošer et al., 2018; Xi et al., 2013). The above prediction models have basically been verified by engineering applications, and they are all suitable for general geological conditions. However, with special geological conditions such as thick unconsolidated layers, they could not guarantee the prediction accuracy of surface subsidence.

Based on the analysis of movement and deformation characteristics of surface monitoring points, an improved Knothe time function model will be proposed. Then a surface subsidence model suitable for mining conditions of thick unconsolidated layers will be given. On the basis, a new mining subsidence prediction model will be established, which aims to provide data support for formulation and implementation of prevention measures of surface damages.

Surface subsidence process in mining areas with thick unconsolidated layers

The process of surface subsidence under thick unconsolidated layers is basically consistent with that under general geological conditions. Considering great difference in lithology between unconsolidated layer and bedrock, the laws of surface movement and deformation are often different from those under general geological and mining conditions.

Surface movement can usually be divided into three stages according to velocity of the maximum subsidence: early, active and decline stage. The early stage starts when the subsidence amount reaches 10 mm, until the subsidence velocity is less than 1.67 mm/d. In the active stage with rapid subsidence, the subsidence velocity is greater than 1.67 mm/d. In the decline stage with stable subsidence, the subsidence velocity is less than 1.67 mm/d, and the accumulated subsidence is not more than 30 mm for 6 consecutive months. Figure 1 shows the change of surface subsidence with time in a mining area with thick unconsolidated layers. It can be seen that the early stage is very short or nonexistent with slow surface deformation. The decline stage was much longer than that of the active stage. The subsidence amount in the active stage accounted for more than 90% of the total subsidence, indicating that the subsidence was more severe in the active stage and the subsidence curve convergence was slower. Moreover, under the condition of thick unconsolidated layers, the subsidence of the center of the surface movement basin was severe, while the boundary converges slowly. After the surface subsidence stopped, the surface damage scope became larger (Ma et al., 2017). Figure 2 shows the fitting results by the probability integral method (PIM).

Figure 1.

Variation of surface subsidence with time.

Figure 2.

Comparison between the fitted value and the measured value of PIM-based prediction model.

The curve edge convergence was slow and PIM was poor in fitting the edge part. This was mainly caused by the surface subsidence due to water loss of thick unconsolidated layers. However, the existing models can’t properly solve the problem (Zhou et al., 2018).

Construction of an improved Knothe time function model

An improved Knothe time function model

In 1952, a Polish scholar Knothe proposed the time function model of surface subsidence. Its function model is as follows:

w (t) = w_{\max} (1 - e^{- c t})

(1)

Where t is the surface subsidence time; w(t) is the subsidence amount of a surface point at time t; w_max is the maximum surface subsidence; c is the time coefficient, related with mechanical properties of the overlying strata.

Liu obtained the first and second derivatives of equation (1) respectively to obtain the surface subsidence velocity and acceleration (Liu, 2010). The change process of subsidence acceleration indicated that the acceleration at the beginning of the surface movement was the maximum negative. It was shown that the subsidence process described by the Knothe time function gradually decreased, which was inconsistent with the dynamic evolution law of surface mining subsidence and movement. The mining-induced surface movement is a complex nonlinear process, and the Knothe function model of surface strata movement is consistent with Kelvin model of creep characteristics (Chen et al., 2017). Therefore, the time function c can be expressed as a power exponent, namely, $c (t) = m t^{k - 1}$ , where are dimensionless model parameters, Then formula (1) can be changed to:

w (t) = w_{\max} (1 - e^{- m t^{k}})

(2)

According to deformation characteristics of surface monitoring points under the condition of thick unconsolidated layers, the water loss of unconsolidated layers will lead to additional surface subsidence whose size is related to the thickness of the unconsolidated layers. The surface subsidence is also affected by the thickness of the bedrock. Therefore, it is proposed to multiply a new coefficient (H/h_s) ⁿ on the basis of formula (2), where H is the mining depth of the coal seam, h_s is the thickness of the unconsolidated layer, n is a dimensionless parameter, and H/h_s is always greater than one. When H is fixed, the larger the h_s, the greater the additional subsidence amount will be. Therefore, the variation degree of the curve can be obtained by adjusting (H/h_s)ⁿ.

Then the improved time function model can be expressed as follows:

w (t) = w_{\max} (1 - e^{- m t^{k}}) {(\frac{H}{h_{s}})}^{n}

(3)

Verification of the model

By equation (3), the first and second derivatives of time t were calculated, respectively. Then subsidence velocity and acceleration can be obtained：

v (t) = w_{1} k m e^{- m t^{k}} t^{k - 1} {(\frac{H}{h_{s}})}^{n}

(4)

a (t) = w_{1} k m e^{- m t^{k}} {(\frac{H}{h_{s}})}^{n} [(k - 1) t^{k - 2} - k m t^{2 (k - 1)}]

(5)

The curves of formulas (3) to (5) were plotted, respectively, as shown in Figures 3 to 5.

It can be seen from Figure 3 that when t = 0, w(0)=0; when t→+∞, w(+∞) = w_max. This showed that the improved model passed through the origin, and there was a maximum subsidence value. It can be seen from Figure 4 that the subsidence velocity was greater than zero, It increased first and then decreased with time. Let a(t) = 0, the following formula could be obtained:

{\begin{array}{l} t_{1} = 0 \\ t_{2} = {(\frac{k - 1}{k m})}^{\frac{1}{k}} \\ t_{3} = \infty \end{array}

(6)

Figure 3.

Single point subsidence curve.

Figure 4.

Single point subsidence velocity curve.

In the phase of t₁→t₂, the subsidence acceleration firstly increased and then decreased, and reached the maximum at t₂; in the t₂→t₃ phase, it was less than zero, and the subsidence velocity gradually decreased to zero. The comparisons among Figures 3 to 5 and the ideal time function curve (Cui et al., 1999) indicated that the subsidence curve, velocity curve and acceleration curve made by the improved Knothe function model basically conformed to the mining-induced surface subsidence process.

Figure 5.

Single point subsidence acceleration curve.

By changing parameters of the dynamic model, the changes of subsidence amount were observed. As shown in Figure 6, m and k had no effect on the initial and final amount of surface subsidence, and only affected the time of reaching the final subsidence. The larger m and k, the shorter the time required to reach stability. n and H/h_s didn’t influence the initial surface subsidence, and only affected the size of the maximum value, and had no effect on the time required for the ground to reach stability. As the value of n increased, the value of surface subsidence gradually increased. As H/hs increased, the maximum subsidence value gradually decreased. This was mainly because as the thickness of the unconsolidated layer decreased, the additional subsidence became smaller.

Figure 6.

Surface subsidence curves with different parameters.

Prediction model of surface subsidence under thick unconsolidated layers

PIM has poor adaptability in mining areas with thick unconsolidated layers. Even with relatively accurate parameters, the subsidence amount obtained by the PIM-based model still has too fast boundary convergence (as shown in Figure 2). In practical applications, although the Boltzmann function model could solve the problem to a certain extent, the fitting accuracy in the entire basin is smaller than that of PIM-based model in some mining areas. On the basis, an improved Boltzmann function model was developed in this paper.

The paramete influence function of Boltzmann function is (Wang et al., 2012)

w_{e} (x) = d W (x) = \frac{1}{R} \frac{\exp (- x / R)}{{(1 + \exp (- x / R))}^{2}}

(7)

Where x is the horizontal distance from the point to the origin of the coordinate, and R is the important radius of influence.

In this paper, the subsidence basin was a combination of two subsidence basins with different parameters R in a certain proportion. Then the prediction formulas of surface movement and deformation were established according to the superposition method, as shown in equation (8):

w_{e} (x) = \frac{1}{R_{1}} \frac{exp (- x / R_{1})}{{(1 + exp (- x / R_{1}))}^{2}} (1- P) + \frac{1}{R_{2}} \frac{exp (- x / R_{2})}{{(1 + exp (- x / R_{2}))}^{2}} P

(8)

Where P is the proportionality coefficient, and R₁ and R₂ are two different important influence radii.

Then the subsidence value caused by semi-infinite mining is as follows:

w (x) = w_{0} \int_{0}^{\infty} \frac{1}{R_{1}} \frac{exp (- (x - s) / R_{1})}{{(1 + exp (- (x - s) / R_{1}))}^{2}} (1- P) + \frac{1}{R_{2}} \frac{exp (- (x - s) / R_{2})}{{(1 + exp (- (x - s) / R_{2}))}^{2}} P d s

(9)

Where w₀ is the maximum subsidence value in the subsidence basin, w₀=mqcosα, m is the thickness of the coal seam, q is the subsidence coefficient, α is the seam dip angle, and s is the horizontal distance from the surface point to the starting point.

After simplification:

w (x) = \frac{w_{0} * (1 - P)}{\exp (- x / R_{1}) + 1} + \frac{w_{0} * P}{\exp (- x / R_{2}) + 1}

(10)

As shown in Figure 7, according to the superposition principle, the subsidence prediction formula of the main section along the strike of the surface moving basin can be obtained as follows:

\begin{array}{l} w^{0} (x) = w (x) - w (x - l_{3}) \\ = m q \cos α \times (\begin{array}{l} [\frac{(1 - P)}{\exp (- x / R_{1}) + 1} + \frac{P}{\exp (- x / R_{2}) + 1}] - \\ [\frac{(1 - P)}{\exp (- (x - l_{3}) / R_{1}) + 1} + \frac{P}{\exp (- (x - l_{3}) / R_{2}) + 1}] \end{array}) \end{array}

(11)

Figure 7.

Movement and deformation calculation principle of main section along the strike.

Figure 8.

Movement and deformation calculation principle of main section along the inclination.

Where l₃ is the calculated length of the finite mining along the strike, l₃=D₃−s₃−s₄.

Similarly, the prediction formula of subsidence on the main section along the inclination can be obtained (the principle is shown in Figure 8), as follows:

\begin{array}{l} w^{0} (y) = w (y) - w (y - l_{1}) \\ = m q \cos α \times (\begin{array}{l} [\frac{(1 - P)}{\exp (- y / R_{1}) + 1} + \frac{P}{\exp (- y / R_{2}) + 1}] - \\ [\frac{(1 - P)}{\exp (- (y - l_{1}) / R_{1}) + 1} + \frac{P}{\exp (- (y - l_{1}) / R_{2}) + 1}] \end{array}) \end{array}

(12)

Where l₁ is the calculated length along the inclination, $l_{1} = (D_{1} - s_{1} - s_{2}) \sin (θ_{0} + α) / \sin (θ_{0})$ ; s₁ and s₂ are offsets of inflection points in downhill and uphill directions; θ₀ is propagation angle of extraction.

In Figure 9, the coal seam was assumed to be a horizontal coal seam. The horizontal projections of tO₁s and xOy coincided. If unit B (s, t) with a width of ds and a length of dt is mined at s, the subsidence value of any point A (x, y) on the surface is as follows:

\begin{array}{l} d w (x, y) = w_{0} w_{e} (x - s) w_{e} (y - t) \\ = w_{0} (\frac{1}{R_{1}} \frac{exp (- (x - s) / R_{1})}{{(1 + exp (- (x - s) / R_{1}))}^{2}} (1- P) + \frac{1}{R_{2}} \frac{exp (- (x - s) / R_{2})}{{(1 + exp (- (x - s) / R_{2}))}^{2}} P) \\ (\frac{1}{R_{1}} \frac{exp (- (y - t) / R_{1})}{{(1 + exp (- (y - t) / R_{1}))}^{2}} (1- P) + \frac{1}{R_{2}} \frac{exp (- (y - t) / R_{2})}{{(1 + exp (- (y - t) / R_{2}))}^{2}} P) \end{array}

(13)

Figure 9.

Schematic diagram of arbitrary point calculation.

In Figure 9, if the strike length of the mining area is D₃ and the inclined length is D₁, the mining-induced surface subsidence value of any point A(x, y) can be expressed as follows:

w (x, y) = w_{0} \int_{0}^{D_{3}} \int_{0}^{D_{1}} d w (x, y)

(14)

w (x, y) = \frac{1}{w_{0}} [w (x) - w (x - D_{3})] [w (y) - w (y - D_{1})]

(15)

According to the derivation of the above deformation prediction formulas and the coal seam dip angle α, equation (15) can be transformed into:

w (x, y) = \frac{1}{w_{0}} [w (x) - w (x - l_{3})] [w (y) - w (y - l_{1})] = \frac{1}{w_{0}} w^{0} (x) w^{0} (y)

(16)

There are nine parameters in the new model: q, P, θ₀, s₁, s₂, s₃, s₄, R₁, R₂. According to the above derivation, the relationship between the predicted subsidence at any point on the surface and the predicted parameters of the subsidence basin can be described as follows:

w_{i} = w (x, y; P_{a}; P_{b})

(17)

Where $P_{a}$ is the parameter matrix of the prediction model, $P_{a} = [q, P, θ_{0}, s_{1}, s_{2}, s_{3}, s_{4}, R_{1}, R_{2}]$ , and $P_{b}$ is the geological and mining condition, such as mining depth, mining height, working face location parameters and coal seam inclination (known conditions).

Assuming that the measured subsidence at any point i is w_si, the estimated subsidence error at this point is:

v_{i} = w_{s i} - w_{i}

(18)

According to the minimum criterion of the sum of squared errors, the constructed criterion for seeking parameters is：

\sum_{i = 1}^{N} v_{i}^{2} = \min

(19)

Where N is the number of measuring points.

Then, the genetic algorithm (Aggestam and Nielsen, 2019; Yarmohammadi et al., 2019) was used to obtain prediction parameters of the model.

Establishment of dynamic prediction model and calculation of parameters

Construction of dynamic prediction model

In this paper, the improved Knothe time function model and the improved Boltzmann time function model were combined. Then, the subsidence $w (x_{i}, y_{i}, t)$ of any point in the whole subsidence basin at any time can be obtained as follows:

w (x_{i}, y_{i}, t) = w_{\max} (1 - e^{- m t^{k}}) {(\frac{H}{h_{s}})}^{n}

(20)

Where $w_{\max}$ is the final subsidence amount of any point, which can be predicted by the improved Boltzmann function model.

Parameter acquisition

The final subsidence amount $w_{\max}$ of the points

Within the influence range of mining subsidence of the working face, the final subsidence amount of any point can be calculated by the improved Boltzmann function prediction model. If the subsidence is not stable, the relevant parameters of the improved Boltzmann function prediction model can be given according to the existing fitting parameters of adjacent or close working faces.

Values of m, k and n

In the improved Knothe time function, m, k and n are related to geological conditions of the working face. They can be obtained by optimization methods (genetic algorithm, particle swarm optimization and etc.), and the fitness function is given, as follows:

{[w_{s i} - w (t, m, k, n)]}^{2} = \min

(21)

Where $w (t, m, k, n)$ is the fitted value.

In the process of case study, it is found that these three parameters are not fixed and are constantly changing. Therefore, they can be fitted according to the measured data to predict the deformation during the next period.

Engineering application

Overview of the study area

This paper took the main section along the inclination of 1414(1) working face of Guqiao Mine in Huainan mining area as an example to establish a dynamic mining subsidence model. The mining of 1414(1) working face started on December 14, 2013, and stopped on November 22, 2014. The fully mechanized mining method was adopted, and all roof caving method was used for roof management. Table 1 shows relevant geological and mining conditions of the working face, and V is the average advance speed.

Table 1.

Geological and mining conditions of 1414 (1) working face.

Working face	H	h_s	α	h	D ₃	D ₁	V
1414(1)	735	411	5	2.7	2120	251	6.18

The layout of monitoring points above 1414 (1) working face is shown in Figure 10. The average interval is 30 m. Many points have been damaged or flooded due to complex environment. In this paper, 4 monitoring points (MS34, MS38, MS46 and MS47) were randomly selected to verify the reliability of the established model. The measured subsidence values are shown in Table 2.

Figure 10.

Location and layout of 1414 working face.

Table 2.

Surface subsidence monitoring data of MS34, MS38, MS46 and MS47.

Observation date	Relative observation	MS34	MS38	MS46	MS47
2013/10/24	0	0	0	0	0
2013/12/22	59	1.8	1.4	2.8	1.3
2014/4/6	164	33.1	32.6	13.7	17
2014/6/22	241	346.4	496.6	74.6	60.6
2014/9/6	317	1011.7	1719.5	497	365
2014/11/22	394	1053.9	1761.9	575.7	434
2015/1/30	463	1072.3	1782.2	585.1	441.4
2015/3/3	495	1100.2	1809.3	607.5	463.6
2015/6/9	593	1105.2	1809	612.9	469.7

Prediction results analysis

Firstly, it is necessary to calculate the maximum subsidence value $w_{\max}$ . The selected working face is the first mining face, and there is no adjacent or similar working face around. The final movement and deformation information has been obtained. Therefore, the method of measured data fitting was applied to get relevant parameters of the improved Boltzmann prediction model of 1414 (1) working face. In order to analyze the predicted performance of the improved Boltzmann model, it was compared with PIM prediction model, and the fitting parameters are shown in Table 3. The fitting values are shown in Figure 11. In Table 3, Er represents the fitting mean square error.

Table 3.

Fitting parameters of the working face.

Models	Parameters
PIM	q	tanβ	θ	s₁	s₂	s₃	s₄	/	/	Er (mm)
PIM	0.94	2.37	88.8	–56	12	–96	–30	/	/	36
Boltzmann	q	R	θ	s₁	s₂	s₃	s₄	/	/	Er/mm
Boltzmann	0.97	76	87.4	56	–98	98	96	/	/	28
Improved Boltzmann	q	P	θ	s₁	s₂	s₃	s₄	R₁	R₂	Er/mm
Improved Boltzmann	1.02	0.45	85	–77	34	–96	–99	54	149	11

Figure 11.

Comparison of fitting data and measured data of 1414 (1) working face.

Table 3 indicated that q and θ changed slightly, and other parameters changed greatly. The fitting accuracy of the improved Boltzmann prediction model was the highest with the mean square error of 11 mm, followed by the Boltzmann prediction model with the mean square error of 28 mm, and the PIM model with the worst fitting accuracy with the mean square error of fitting of 36 mm. The overall fitting accuracy of the improved Boltzmann function model was higher, and the subsidence values at the boundary were in good agreement.

The dynamic parameters of the working face were obtained according to the method in Section Parameter acquisition. The fitting parameters and errors are shown in Table 4. The comparison between the fitting values and the measured values is shown in Figure 12. In Table 4, Er1 and Er2 represent the fitting mean square error of the improved Knothe function and the Knothe function, respectively.

Table 4.

Dynamic prediction parameters and fitting mean square errors of four points.

Monitoring points	m	k	n	Er1 (mm)	Er2 (mm)
MS34	4.637e-18	7.104	-0.289	13.295	244.612
MS38	4.399e-21	8.341	-0.018	13.635	428.222
MS46	4.571e-24	9.434	-0.050	11.001	148.268
MS47	1.681e-22	8.792	-0.065	11.081	109.123

Figure 12.

Comparison of two time-function models for surface subsidence prediction: (a) MS34, (b) MS38, (c) MS46, (d) MS47.

Table 4 and Figure 12 showed that the prediction accuracy of the improved Knothe time function model was higher than that of Knothe time function model. The improved Knothe model could better reflect the process of surface subsidence, and the fitting results were in good agreement with the measured results. It can provide some reference for dynamic prediction of surface subsidence and movement in mining areas with thick unconsolidated layers.

Conclusions

In order to overcome the deficiency of the Knothe time function, its influence coefficient was changed into coefficient in exponential form. Moreover, the influence coefficient ${(H / h_{s})}^{n}$ of the unconsolidated layer was added. The subsidence amount, subsidence velocity and subsidence acceleration described by the improved model were all consistent with the actual situation.

A prediction model suitable for the surface subsidence basin of thick unconsolidated layers was established. Then a new mining subsidence dynamic prediction model was developed by combining this model with the improved Knothe time function model.

The measured data of 1414(1) working face of Huainan mining area were used to verify the rationality of the model. In the 4 monitoring points, the fitting mean square errors of the improved Knothe model were between 11.001 and 13.635 mm, and the fitting mean square errors of the Knothe model were between 109.123 and 428.222 mm. The fitting results show that the improved model can better reflect the dynamic process of mining subsidence, and the prediction data and the measured data were in good agreement.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research is sponsored by the funding from National Natural Science Foundation of China (Grant Nos. 41474026, 52074010, 41602357), and the Natural Science Foundation of Anhui Province, China (Grant Nos.2008085MD114, 2008085QD178).

ORCID iD

Shenshen Chi

References

Aggestam

Nielsen

JCO

(2019) Multi-objective optimisation of transition zones between slab track and ballasted track using a genetic algorithm. Journal of Sound and Vibration 446: 91–112.

Chen

Niu

Liu

et al;. (2017) Improvement of Knothe model and analysis on dynamic evolution law of strata movement in fill mining. Journal of China University of Mining and Technology 46(2): 250–256.

Cui

Liao

Zhao

et al;. (1999) Discussion on the time function of time dependent surface movement. Journal of China Coal Society 5: 453–456.

Djamaluddin

Mitani

Esaki

(2011) Evaluation of ground movement and damage to structures from Chinese coal mining using a new GIS coupling model. International Journal of Rock Mechanics and Mining Sciences 48(3): 380–393.

Kowalski

(2001) Surface subsidence and rate of its increments based on measurements and theory. Archives of Mining Sciences 46(4): 391–406.

Zha

Guo

(2019) A new dynamic prediction method for surface subsidence based on numerical model parameter sensitivity. Journal of Cleaner Production 233: 1418–1424.

Liu

(2010) Study on the dynamic course of the surface subsidence and the model based on theory of key rock stratum. PhD Thesis, Chongqing University, China.

Zhang

(2017) Subsidence prediction method of solid backfilling mining with different filling ratios under thick unconsolidated layers. Arabian Journal of Geoences 10(23): 511.

Nie

Wang

et al;. (2015) A new prediction model for mining subsidence deformation: The arc tangent function model. Natural Hazards 75(3): 2185–2198.

10.

Roser

Potočnik

Vulić

(2018) Analysis of dynamic surface subsidence at the underground coal mining site in Velenje. Slovenia through modified sigmoidal function. Minerals 8(2): 74.

11.

Unlu

Akcin

Yilmaz

(2013) An integrated approach for the prediction of subsidence for coal mining basins. Engineering Geology 166: 186–203.

12.

Wang

Xuan

(2018) Time function model of dynamic surface subsidence assessment of grout-injected overburden of a coal mine. International Journal of Rock Mechanics and Mining Sciences 104: 1–8.

13.

Wang

Liu

et al;. (2012) Study on prediction of surface subsidence in mined-out region with the MMF model. Journal of China Coal Society 37(3): 411–415.

14.

Wang

Liu

et al;. (2012) Model for mining subsidence prediction based on Boltzmann function. China Coal Society 38(8): 1352–1356.

15.

Hong

Shao

(2013) Application of improved logistic function model to prediction of surface subsidence. Coal Science and Technology 41: 114–117.

16.

Yang

Zhu

et al;. (2017) Deriving dynamic subsidence of coal mining areas using InSAR and logistic model. Remote Sensing 9(2): 125–144.

17.

Yarmohammadi

Rafiee-Dehkharghani

Behnia

et al;. (2019) Design of wave barriers for mitigation of train induced vibrations using a coupled genetic-algorithm/finite-element methodology. Soil Dynamics and Earthquake Engineering 121: 262–275.

18.

Dang

Pan

et al;. (2003) Theological characteristics of surface dynamic subsidence by mining. Journal of Xi'an University of Science and Technology 2: 131–134.

19.

Zhang

Cheng

Yao

et al;. (2020) Application of the improved Knothe time function model in the prediction of ground mining subsidence: a case study from Heze City, Shandong province, China. Applied Sciences 10(9): 3147.

20.

Zhao

(2005) Rheological model and its application in dynamic prediction of mining subsidence. Sci-Tech Information Development and Economy 5: 195–196.

21.

Zhou

Miao

et al;. (2018) Combined prediction model for mining subsidence in coal mining areas covered with thick alluvial soil layer. Bulletin of Engineering Geology and the Environment 77(1): 283–304.

Research on dynamic prediction model of surface subsidence in mining areas with thick unconsolidated layers

Abstract

Keywords

Introduction

Surface subsidence process in mining areas with thick unconsolidated layers

Construction of an improved Knothe time function model

An improved Knothe time function model

Verification of the model

Prediction model of surface subsidence under thick unconsolidated layers

Establishment of dynamic prediction model and calculation of parameters

Construction of dynamic prediction model

Parameter acquisition

The final subsidence amount w max ⁡ of the points

Values of m, k and n

Engineering application

Overview of the study area

Prediction results analysis

Conclusions

Footnotes

Declaration of conflicting interests

Funding

ORCID iD

References

The final subsidence amount $w_{\max}$ of the points