Failure mechanism and control technology for the surrounding rock of retreat roadways beneath remnant coal pillars in close-distance coal seams Authors: Yang Wang, Zhiqiang Zhao, Hongtao Liu, Jinwen Hao, Xiaoxiang Wei Published: 2026 ABSTRACT This study takes the 13,402 return airway of Sunjiagou Coal Mine as the engineering background and investigates the failure and control of surrounding rock in response to asymmetric loading and deformation instability of an underlying mining roadway affected by a remnant coal pillar in closely spaced coal seams. A combination of theoretical analysis, FLAC3D numerical simulation, borehole observation, and field industrial testing was adopted to analyze stress transfer, plastic-zone evolution, and surrounding-rock failure characteristics within the remnant coal pillar–floor–roadway surrounding-rock system. The results show that roadway stability is improved when the roadway is arranged with an outward offset or directly beneath the coal pillar. When the coal pillar width is 25–30 m and the inter burden thickness is no less than 30 m, the disturbance induced by the overlying remnant coal pillar on the underlying roadway is significantly weakened. Roof fractures are mainly developed within the range of 0–5 m. Based on these findings, an optimized support scheme involving coordinated long and short roof cables and targeted reinforcement of the roadway ribs was proposed. Field tests indicate that the deformation of the roof, floor, and both ribs is controlled within 4–5 mm. The research results can provide a theoretical basis and engineering reference for roadway layout and support design in mining roadways affected by remnant coal pillars in closely spaced coal seams. Introduction As coal resource extraction in China progressively shifts toward greater depths and more complex geological conditions, the efficient and safe mining of close-distance coal seam groups has become a major challenge in mining engineering. In close-distance coal seam mining, the asymmetric stress field induced by remnant coal pillars in the upper seam significantly affects the stability of the surrounding rock of roadways in the underlying seam, and can readily trigger hazards such as roadway deformation, roof falls, and rock bursts. In particular, because remnant coal pillars act as concentrated load-transfer carriers for the overburden, the complexity of stress re-distribution in the underlying strata and the associated failure mechanisms of the surrounding rock has become a key scientific issue constraining safe mining. Owing to its notable advantages in reducing coal resource loss and improving recovery ratio, gob-side entry driving has become a core technique for efficient mining of close-distance coal seam groups in China and has been widely applied in major mining areas1–3. However, when a roadway in the lower seam lies within the influence zone of a remnant coal pillar in the upper seam, the pillar—acting as a concentrated load-transfer carrier for the overburden—can generate an asymmetric high-stress region beneath it; coupled with the dual disturbances of roadway excavation and extraction of an adjacent working face, this leads to pronounced stress redistribution in the surrounding rock and readily induces hazards such as rib heave, roof subsidence, and even rock bursts4–6. The return-air roadway of the 13,402 working face in the Sunjiagou Coal Mine, Shanxi Province, faces such problems. The roadway has been subjected to the long-term influence of remnant coal pillars in the overlying No. 11 coal seam, and the existing support scheme, lacking a systematic understanding of pillar stress distribution and the surrounding-rock failure mechanism, is difficult to adapt to the deformation demands under an asymmetric stress environment; support failures are therefore frequently observed on site, severely constraining mining efficiency and construction safety7–9. Numerous studies have been conducted by researchers worldwide on coal pillars and the stability of gob-side roadways. In international research, Wilson’s “two-zone confinement theory” first clarified the strength difference between the “nucleus zone (load-bearing core)” and the “nonnucleus zone (marginal crushed zone)” of a coal pillar, revealing the intrinsic relationship between pillar dimensions and load-bearing capacity and providing a classical theoretical framework for rational pillar design10. Based on field measurements from multiple mining areas in South Africa, Salamon developed an empirical coal-pillar strength formula that quantified the effects of pillar width and height on its load-bearing limit; this formula remains an important reference for coal-pillar design11. Drawing on longwall mining practice, Whittaker and Singh established models for coal-pillar design and stability assessment and clarified the relative influence of pillar width and seam thickness on stability12. Domestic research is more closely aligned with the mining conditions of close-distance thick coal seam groups in China. Through theoretical analysis and field measurements, Zhang et al. systematically revealed the evolutionary pattern of “mining disturbance–stress redistribution–deformation and failure” of the surrounding rock in extremely close-distance coal seam mining, providing a theoretical basis for subsequent development of surrounding-rock control technologies13. Using a combined approach of laboratory similarity simulation and in-situ field monitoring, Zhang Jian et al. accurately captured the transfer path and attenuation law of concentrated stress in coal pillars within close-distance coal seams, providing experimental support for the optimization of pillar parameters14. Li et al. further investigated the failure characteristics of mining roadways affected by stress accumulation from residual pillars in contiguous coal seams and proposed corresponding control techniques, thereby highlighting the critical role of residual-pillar-induced stress concentration in roadway instability15. Chen et al. investigated the failure mechanism of borehole stress-gauge monitoring under coal seam mining conditions and proposed a new in-situ monitoring method for mining stress, providing useful support for identifying stress evolution and surrounding-rock response under complex mining conditions16. In terms of roadway layout and support technologies, Kang Guanxian, focusing on extremely close-distance coal seams beneath goafs, established a correlation model between the reasonable location of retreat roadways and the distance to the goaf as well as the inter-seam spacing, and proposed a stress-avoidance roadway layout scheme17. Li Guan et al. focused on the influence of residual-section coal pillars and optimized the relative position between the roadway and the pillar through numerical simulation, reducing roadway maintenance costs by 25%18. Liu Hongtao, considering the characteristics of downward mining, analyzed the dynamic evolution of the stress field and established a dynamic optimization model for roadway positioning19. Wang et al. further investigated optimized positioning and cross-layer control for roadways beneath residual coal pillars in extremely close-distance coal seams, providing new references for roadway layout and hierarchical surrounding-rock control under strong mining disturbance20. Chen et al. studied the reasonable location of the stopping line in close-distance underlying coal seams and the partition support of large cross-section roadways, providing useful references for roadway layout and surrounding-rock control under mining disturbance21. Chen et al. examined the failure mechanism and divisional differentiated control of surrounding rock in mining roadways under remaining coal pillars in close-distance coal seams, demonstrating that targeted support should be designed according to the spatial distribution of damage and stress concentration22. In the field of support technologies, the team led by Kang Hongpu developed the CRMG700 ultra-high-strength rock bolt and established a synergistic control technique of “support–surrounding-rock modification–stress relief”, achieving a reduction of more than 40% in roadway deformation in deep mine applications23. The constant-resistance large-deformation bolt technology proposed by He Manchao effectively mitigates asymmetric deformation in high-stress roadways by absorbing deformation energy of the surrounding rock24. Du Huailong et al. innovatively exploited the stress-disturbance characteristics of overlying remnant coal pillars to optimize the mining sequence and support parameters, achieving an 8% increase in resource recovery ratio in a mine in Huainan, Anhui Province25. In recent years, related studies have further deepened the research on surrounding-rock control in closely spaced coal seams from the perspectives of the rational location of the stopping line, zonal support for large-section roadways, the failure mechanism of stress monitoring in coal seam mining, and new monitoring methods. These studies provide new insights into roadway layout optimization and monitoring evaluation under complex stress environments. These findings provide important support for understanding the mechanical en-vironment of roadways beneath coal pillars, yet limitations remain. International studies are mostly based on geological conditions characterized by thin seams, gentle dip angles, and shallow mining depths, and their conclusions are difficult to directly adapt to the complex engineering scenarios in China involving close-distance thick seam groups and large mining depths. Domestic studies, although closer to practical conditions, remain insufficient in analyzing the coupled influences of multiple fac-tors—such as coal-pillar size, inter-seam spacing, and roadway offset distance—on the evolution of surrounding-rock stress and plastic zones under the specific conditions of the Sunjiagou Coal Mine, including the spacing between the No. 11 and No. 13 seams and the dimensions of remnant pillars in the No. 11 seam. Moreover, targeted sup-port-optimization studies for the 13,402 return-air roadway are still lacking, making it difficult to provide direct guidance for field construction. Based on the specific conditions of the 13,402 return airway in Sunjiagou Coal Mine, this study analyzes the coupled effects of roadway offset distance, coal pillar width, and interburden thickness on the response of the surrounding rock. The numerical analysis results are further integrated with field support optimization, aiming to provide a basis for surrounding-rock control in roadways under similar conditions. Engineering background Mine overview This study takes the return airway of the 13,402 working face in the No. 13 coal seam of Sunjiagou Coal Mine, Shanxi Province, as the engineering background. Sunjiagou Coal Mine is currently mining the No. 13 coal seam, and the 13,402 working face is a successive working face. The originally exposed total thickness of the No. 13 coal seam in the 13,402 working face is 14.06 m, including a dirt-band thickness of 0.7 m. To facilitate the simplified stratified treatment in the subsequent numerical simulation, the dirt band was equivalently incorporated into the adjacent surrounding rock in the model. Therefore, the effective coal seam thickness adopted in the numerical model is 13.4 m. The roof is mainly composed of black mudstone and dark-gray sandy mudstone, while the floor mainly consists of gray medium sandstone and black mudstone. The coal seam generally occurs as a monoclinic structure, dipping from east to west, with a dip angle of 5°–8° and an average of 6°. The 13,402 return airway is located beneath the goafs of the 11,202 and 11,203 working faces and is affected by the remnant coal pillar in the overlying No. 11 coal seam, as shown in Fig. 1. Under the combined influence of closely spaced coal seam mining and the overlying remnant coal pillar, the surrounding rock of the 13,402 return airway is subjected to a pronounced asymmetric stress environment. After the superposition of stress transfer from the remnant coal pillar and excavation-induced disturbance, the roof and both ribs of the roadway become more susceptible to deformation and instability. The existing conventional support parameters are therefore difficult to adapt effectively to the requirements of surrounding-rock control under such a stress environment. Accordingly, investigating the failure mechanism of the surrounding rock in a mining roadway beneath a remnant coal pillar, and further optimizing the roadway layout and support parameters on this basis, is of clear practical engineering significance. Fig. 1Mining relationship and rock layer histograms. Mining relationship and rock layer histograms. The hydrogeological conditions of the 13,402 working face are moderate; the main aquifers are the water-bearing strata of the Carboniferous Taiyuan Formation and the floor aquifer of the Shanxi Formation, and the principal water hazards include fissure water, roof water, floor water, confined water, and accumulated water in the goaf. During roadway development, the expected main forms of water inflow include localized roof “dripping” and floor seepage/inflow. Influenced by the regional tectonic setting, the Sunjiagou Coal Mine field is generally characterized by a monocline with an approximately north–south strike and a westward dip. Most of the area is covered by loess, and the strata dip angles range from 3° to 9°. Through 3D seismic exploration and roadway exposures at the 11,201, 11,203, 11,204, and 11,106 working faces, a total of 34 minor faults with throws of 0–5 m and one collapse column were identified. No folds or igneous intrusions were observed. Roadway layout of the working face The 13,402 working face is arranged with two roadways, namely the intake-air roadway and the return-air roadway. The remnant coal pillar is currently located above the return-air roadway; therefore, this study focuses only on the return-air roadway, as shown in Fig. 1. At present, the roof of the 13,402 return-air roadway is supported by an all-cable-bolt system, and the roadway cross-section and support parameters are listed in Table 1. Table 113,402 return-air roadway cross-section and support parameters.Roadway nameRoadway cross-sectional shapeLength / mWidth / mWidth × height / mSupport typeReturn-air roadwayRectangular cross-section4.83.64.8 × 3.6Rock bolts + cable bolts + reinforcing mesh + steel straps Table 113,402 return-air roadway cross-section and support parameters.Roadway nameRoadway cross-sectional shapeLength / mWidth / mWidth × height / mSupport typeReturn-air roadwayRectangular cross-section4.83.64.8 × 3.6Rock bolts + cable bolts + reinforcing mesh + steel straps 13,402 return-air roadway cross-section and support parameters. During roadway development, the support design parameters are as follows: (1) Roadway roof. Short cable bolts: The roadway roof is supported by six Φ21.8 × 5200 mm high-strength, low-relaxation prestressed steel-strand cable bolts (1 × 19 construction), arranged at 800 mm spacing and 1000 mm row spacing. Long cable bolts: The roadway roof is supported by two Φ21.8 × 8200 mm high-strength, low-relaxation prestressed steel-strand cable bolts (1 × 19 construction), arranged at 2200 mm spacing and 1000 mm row spacing. Fig. 2Schematic diagram of support layout during roadway excavation (a) Roadway support cross-sectional view (b) Roadway support plan view. Schematic diagram of support layout during roadway excavation (a) Roadway support cross-sectional view (b) Roadway support plan view. (2) Roadway ribs. On the coal-pillar side of the roadway, Φ22 × 2400 mm ribless threaded-steel resin bolts are installed with a spacing and row spacing of 1000 mm × 1000 mm, supplemented by two Φ21.8 × 8200 mm high-strength, low-relaxation prestressed steel-strand cable bolts (1 × 19 construction) with a spacing of 2000 mm and a row spacing of 1000 mm. On the mining side of the roadway, Φ22 × 2400 mm ribless threaded-steel resin bolts are installed with a spacing and row spacing of 1000 mm × 1000 mm. The support design parameters for roadway development are shown in Fig. 2. Field measurements of roof structural characteristics in the 13,402 return-air roadway A borehole endoscope was used to observe the surrounding-rock structure above the roadway roof. The endoscopic probe was placed into a borehole drilled in the roadway roof, and the endoscopic camera was used to acquire and process information on the surrounding rock around the borehole, generate a video signal, and display the images on the monitor of the endoscope computer host, thereby achieving endoscopic detection of the roadway roof. Endoscopic observations were conducted at different locations with respect to the roadway development face. The endoscopic data were exported for analysis, and bar charts were produced using bar-plot software. The results are shown in Fig. 3. Fig. 3Borehole endoscopic images. Borehole endoscopic images. As indicated by the borehole endoscopic observations in Fig. 3, roof fractures and damage are also concentrated within the 2.0–5.0 m range. An annular crack was observed on the borehole wall at 0.75 m, a longitudinal crack occurred at 1.87 m, slight fragmentation was present at 3.52 m, and further breakage was observed at 5.04 m. The overall analysis indicates that, based on endoscopic observations at different locations relative to the development face of the 13,402 return-air roadway, roof fracturing is mainly within the 0–5 m range. The roadway roof is relatively intact, with only a small number of cracks in the shallow zone. Borehole observation results indicate that roof fractures and local fragmentation in the 13,402 return airway are mainly developed within the range of 0–5 m, suggesting that the shallow-to-intermediate roof surrounding rock has been significantly disturbed. This finding is consistent with the plastic-zone distribution characteristics revealed by the subsequent numerical simulation, and provides a basis for identifying key reinforcement zones and optimizing support parameters for the roadway. Study on the stress distribution of surrounding rock in roadways beneath remnant coal pillars Load-bearing analysis of the coal pillar The load borne by the remnant coal pillar mainly consists of two components: (i) the vertical pressure P2 transmitted from the overlying strata, and (ii) the interaction force exerted by key block B in a hinged state. To simplify the analysis, a unit width of the remnant coal pillar is considered; the concentrated stress applied by the hinged key block on the pillar is therefore treated as a concentrated force, and the peak stress of the coal pillar, σ1is given by:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\sigma _1}={q_0}+{Q_A}={k_1}\gamma H$$\end{document} According to the limit equilibrium zone theory26, the width of the plastic zone of the remnant coal pillar 𝑏 :2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$b=\frac{{\lambda m}}{{2\tan \varphi }}\ln \left( {\frac{{{\sigma _1}+\frac{c}{{\tan \varphi }}}}{{\frac{c}{{\tan \varphi }}+\frac{p}{\lambda }}}} \right)$$\end{document} Abutment pressure in the elastic pressurization zone:3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sigma ={\sigma _1}\exp \left( { - \frac{f}{{m\lambda }}\frac{{1 - \sin \varphi }}{{1+\sin \varphi }}} \right)$$\end{document} When σ = γH:4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d=\frac{{m\lambda }}{f} \cdot \frac{{1+\sin \varphi }}{{1 - \sin \varphi }}\ln {k_1}$$\end{document} In Eqs. (1–4), q0 is the in-situ stress, λ the lateral pressure coefficient, k1 the peak stress concentration factor, d the extent of the pillar’s elastic pressurization zone, γ the weighted average unit weight of the overburden, m the mining height, c the coal cohesion, p the rib support intensity, H the vertical burial depth, f the interlayer friction coefficient, and φ the coal internal friction angle. It is assumed that the influence range of the lateral abutment pressure, S, can be expressed as the sum of b and d. By analyzing the relationship between the protective pillar width L and the abutment-pressure influence range S, the stress state of the coal pillar can be classified into three representative distribution patterns: When the coal pillar width L is significantly greater than twice the influence range (L > 2 S), the central part of the pillar remains at the in-situ stress level; When the coal pillar width falls between the influence range and twice the influence range (S < L<2 S), the abutment pressures from both sides superimpose in the middle of the pillar, forming a characteristic saddle-shaped stress distribution; When the coal pillar width is smaller than the influence range (L < S), the peak-stress zones on both sides overlap, ultimately forming a single stress peak at the center of the pillar. Theoretical calculation of coal-pillar load capacity Table 2Design parameters for coal pillar load-bearing capacity calculation.Calculation parameterMaximum fractured-zone height / mLoad-bearing strata thickness above key stratum / mAverage unit weight of overburden kN/m3Average unit weight of key stratum kN/m3Key-block length on both sides / mValue4025242518.5Calculation parameterBlock rotation angle (°)Lateral pressure coefficient / λInterlayer friction coefficient/ µDesigned coal-pillar width / mMain-roof key-stratum thickness / mValue40.640.231513.8 Table 2Design parameters for coal pillar load-bearing capacity calculation.Calculation parameterMaximum fractured-zone height / mLoad-bearing strata thickness above key stratum / mAverage unit weight of overburden kN/m3Average unit weight of key stratum kN/m3Key-block length on both sides / mValue4025242518.5Calculation parameterBlock rotation angle (°)Lateral pressure coefficient / λInterlayer friction coefficient/ µDesigned coal-pillar width / mMain-roof key-stratum thickness / mValue40.640.231513.8 Design parameters for coal pillar load-bearing capacity calculation. Interlayer friction coefficient / µ The mining parameters for the No. 11 coal seam in Sunjiagou Coal Mine are as follows: the seam occurs at a depth of 280 m, with an average mining height of 2.5 m. Tests of coal mechanical properties indicate a cohesion of 1.9 MPa, and the designed support intensity of the coal-rib support system is 0.08 MPa. The remaining calculation parameters are listed in Table 2. Based on engineering mechanics principles, the load of the overlying strata on the coal pillar is simplified as the in-situ stress; using a uniformly distributed load model, the in-situ stress is calculated to be 6.8 MPa. Based on the above theoretical model Eqs. (1–4) and the determined parameter set, numerical substitution yields: The overlying load on key block B, P1:5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${P_1}={L_1}{\gamma _1}{H_1}=18.5 \times 24 \times 25=11100kN$$\end{document} The self-weight (action force) of key block B, PB:6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${P_B}={L_1}{\gamma _2}{h_1}=18.5 \times 25 \times 13.8=6382.5kN$$\end{document} The force exerted by key block B on the coal pillar, QA:7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${Q_A}=\frac{{4{h_1} - 3{L_1}sin\theta }}{{2(2{h_1} - {L_1}\sin \theta )}}({P_1}+{P_B})=17482.5kN$$\end{document} The peak stress of the coal pillar, σ1:8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\sigma _1}=\gamma H+{Q_A}=24.1MPa \approx 3.44\gamma H$$\end{document} The plastic-zone width of the coal pillar, b:9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$b=\frac{{\lambda m}}{{2\tan \varphi }}\ln \left( {\frac{{k\gamma H+\frac{c}{{\tan \varphi }}}}{{\frac{c}{{\tan \varphi }}+\frac{p}{\lambda }}}} \right)={\mathrm{3}}{\mathrm{.2}}m$$\end{document} The extent of the elastic pressurization zone, d:10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d=\frac{{m\lambda }}{f}\frac{{1+\sin \varphi }}{{1 - \sin \varphi }}\ln {k_1}=8.{\mathrm{3}}m$$\end{document} The influence range of the abutment pressure, S:11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$0.5L{\text{ }}<{\text{ }}S=b{\text{ }}+{\text{ }}d=3.2+8.3{\text{ }}={\text{ 11}}.5m{\text{ }}<{\text{ }}L$$\end{document} The calculated results indicate pronounced stress superposition in the central region of the coal pillar. A mechanical analysis of Eq. (3) shows that, after accounting for the superposition of stresses from both sides, the resultant stress at this location, σ2, increases to approximately 10.7 MPa. Stress distribution of the floor beneath the remnant coal pillar in the no. 11 coal seam A stress-distribution model for the floor beneath the remnant coal pillar in the no. 11 coal seam After mining of the upper coal seam, the remnant coal pillar and the hinged key block jointly form a composite load-bearing system that transfers the load of the overlying strata. This load-bearing system drives the stress within the coal pillar to be transmitted to the floor strata, thereby causing a redistribution of the floor stress field. Based on an in-depth analysis of the cooperative load-bearing mechanism of the coal pillar–hinged key block system, and considering that the self-weight of caved gangue in the goaf is much smaller than the overburden load and can be neglected, the following mechanical model can be constructed: the initial load on the floor within the goaf on both sides of the coal pillar can be approximately taken as zero. As the caved gangue in the goaf is gradually compacted, the load borne by the floor strata increases progressively and ultimately approaches the in-situ stress state. To evaluate the stress influence of the remnant coal pillar on the floor strata, a stress-calculation model for the floor beneath the pillar can be established based on the elastic half-plane theory27, taking the center of the upper boundary of the pillar–floor interface as the coordinate origin, as shown in Fig. 4. Fig. 4Stress calculation model for the floor stratum. Stress calculation model for the floor stratum. Based on the interaction between the coal pillar and the floor, an analytical expression is derived to describe the stress distribution along the upper boundary of the floor strata, as follows:12\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${q_1}=\frac{\xi }{d}({\sigma _1} - {\sigma _2})+{\sigma _0}{q_1}=\frac{\xi }{d}({\sigma _1} - {\sigma _2})+{\sigma _0}$$\end{document}13\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${q_2}= - \frac{\xi }{d}({\sigma _1} - {\sigma _2})+{\sigma _0}$$\end{document}14\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${q_3}= - \frac{{{\sigma _1}}}{b}\left( {\xi - \frac{L}{2}} \right)$$\end{document}15\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${q_4}=\frac{{{\sigma _1}}}{b}\left( {\xi +\frac{L}{2}} \right)$$\end{document} Through an in-depth analysis of the stress-field distribution characteristics in the pressure-relief zone, a mechanical calculation model for the equivalent triangular load on the floor is established as follows:16\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${q_5}=\frac{{{q_0}}}{e}\left( {\xi - \frac{L}{2}} \right)$$\end{document}17\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${q_6}= - \frac{{{q_0}}}{e}\left( {\xi +\frac{L}{2}} \right)$$\end{document}18\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left\{ {\begin{array}{*{20}{l}} {d{\sigma _x}= - \frac{{2qd\xi }}{\pi }\frac{{{y^3}}}{{{{[{{(x - \xi )}^2}+{y^2}]}^2}}}} \\ {d{\sigma _y}= - \frac{{2qd\xi }}{\pi }\frac{{{{(x - \xi )}^2}y}}{{{{[{{(x - \xi )}^2}+{y^2}]}^2}}}} \\ {d{\tau _{xy}}= - \frac{{2qd\xi }}{\pi }\frac{{(x - \xi ){y^2}}}{{{{[{{(x - \xi )}^2}+{y^2}]}^2}}}} \end{array}} \right.$$\end{document} According to the superposition principle in elasticity, the stress state at any point within the floor strata can be determined by linearly combining the stress fields generated by each load component acting individually. Given the symmetry of this mechanical system about the coal-pillar centerline, the analytical stress solutions corresponding to each symmetric load component share a unified mathematical form. For brevity, only the analytical solution of the stress field induced by unilateral loading with respect to the coal-pillar centerline is presented as follows: By substituting the functional expression of ξ into Eq. (11) and then performing the integration, the floor stress components induced by the load q1 can be obtained as follows:19\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} {\sigma _{{x_1}}}= - \frac{2}{\pi }\int_{0}^{d} {\frac{{{q_1}{y^3}d\xi }}{{{{[{{(x - \xi )}^2}+{y^2}]}^2}}}} =\frac{{({\sigma _1} - {\sigma _0})y}}{{\pi d}}\left[ {\frac{1}{{{{(x - d)}^2}+{y^2}}} - \frac{1}{{{x^2}+{y^2}}}} \right] \hfill \\ - \frac{{({\sigma _1} - {\sigma _0})x+{\sigma _0}d}}{{\pi d}} \times \left[ {\arctan \frac{x}{y} - \arctan \frac{{x - d}}{y}+\frac{{xy}}{{{x^2}+{y^2}}} - \frac{{(x - d)y}}{{{{(x - d)}^2}+{y^2}}}} \right] \hfill \\ \end{gathered}$$\end{document}20\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} {\sigma _{{y_1}}}= - \frac{2}{\pi }\int_{0}^{d} {\frac{{{q_1}{{(x - \xi )}^2}yd\xi }}{{{{[{{(x - \xi )}^2}+{y^2}]}^2}}}} = - \frac{{({\sigma _1} - {\sigma _0})y}}{{\pi d}}\left[ {\ln \left| {\frac{{{{(x - d)}^2}+{y^2}}}{{{x^2}+{y^2}}}} \right| - \left( {\frac{{{y^2}}}{{{{(x - d)}^2}+{y^2}}} - \frac{{{y^2}}}{{{x^2}+{y^2}}}} \right)} \right] \hfill \\ - \frac{{({\sigma _1} - {\sigma _0})x+{\sigma _0}d}}{{\pi d}}\left[ {\arctan \frac{x}{y} - \arctan \frac{{x - d}}{y} - \frac{{xy}}{{{x^2}+{y^2}}}+\frac{{(x - d)y}}{{{{(x - d)}^2}+{y^2}}}} \right] \hfill \\ \end{gathered}$$\end{document}21\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} {\tau _{x{y_1}}}= - \frac{2}{\pi }\int_{0}^{d} {\frac{{{q_1}(x - \xi ){y^2}d\xi }}{{{{[{{(x - \xi )}^2}+{y^2}]}^2}}}} =\frac{{({\sigma _1} - {\sigma _0})y}}{{\pi d}}\left[ {\arctan \frac{x}{y} - \arctan \frac{{x - d}}{y} - \frac{{xy}}{{{x^2}+{y^2}}}+\frac{{(x - d)y}}{{{{(x - d)}^2}+{y^2}}}} \right] \hfill \\ +\frac{{({\sigma _1} - {\sigma _0})x+{\sigma _0}d}}{{\pi d}}\left[ {\frac{{{y^2}}}{{{x^2}+{y^2}}} - \frac{{{y^2}}}{{{{(x - d)}^2}+{y^2}}}} \right] \hfill \\ \end{gathered}$$\end{document} Floor stress components induced by the load q3:22\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} {\sigma _{x3}}= - \frac{2}{\pi }\int_{d}^{{\frac{L}{2}}} {\frac{{{q_3}{y^3}d\xi }}{{{{[{{(x - \xi )}^2}+{y^2}]}^2}}}} \hfill \\ = - \frac{{{\sigma _1}}}{{\pi b}}\left[ {\frac{{\left( {x - \frac{L}{2}} \right)(x - d)y+{y^3}}}{{{{(x - d)}^2}+{y^2}}} - \frac{{4{y^3}+2y{{(2x - L)}^2}}}{{{{(2x - L)}^2}+4{y^2}}}+\left( {x - \frac{L}{2}} \right)\left( {\arctan \frac{{x - d}}{y} - \arctan \frac{{2x - L}}{{2y}}} \right)} \right] \hfill \\ \end{gathered}$$\end{document}23\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} {\sigma _{y3}}= - \frac{2}{\pi }\int_{d}^{{\frac{L}{2}}} {\frac{{{{(x - \xi )}^2}{q_3}yd\xi }}{{{{[{{(x - \xi )}^2}+{y^2}]}^2}}}} \hfill \\ =\frac{{{\sigma _1}}}{{\pi b}}\left\{ \begin{gathered} \left( {x - \frac{L}{2}} \right)\left( {\arctan \frac{{x - d}}{y} - \arctan \frac{{2x - L}}{{2y}}} \right) - \frac{{(2x - L)(x - d)y}}{{2{{(x - d)}^2}+2{y^2}}}+ \hfill \\ \frac{{{{(2x - L)}^2}y}}{{{{(2x - L)}^2}+4{y^2}}}+y\left[ {\ln \left| {{{\left( {x - \frac{L}{2}} \right)}^2}+{y^2}} \right| - \ln \left| {{{(x - d)}^2}+{y^2}} \right|} \right] - \hfill \\ \frac{{4{y^3}}}{{{{(2x - L)}^2}+4{y^2}}}+\frac{{{y^3}}}{{{{(x - d)}^2}+{y^2}}} \hfill \\ \end{gathered} \right\} \hfill \\ \end{gathered}$$\end{document}24\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} {\tau _{x{y_3}}}= - \frac{2}{\pi }\int_{d}^{{\frac{L}{2}}} {\frac{{{q_2}(x - \xi ){y^2}d\xi }}{{{{[{{(x - \xi )}^2}+{y^2}]}^2}}}} \hfill \\ = - \frac{{{\sigma _1}}}{{\pi b}}\left[ \begin{gathered} y\left( {\arctan \frac{{x - d}}{y} - \arctan \frac{{2x - L}}{{2y}}} \right) \hfill \\ - \frac{{(x - d - 1){y^2}}}{{{{(x - d)}^2}+{y^2}}}+\frac{{(4x - 2L - 1){y^2}}}{{{{(2x - L)}^2}+4{y^2}}} \hfill \\ \end{gathered} \right] \hfill \\ \end{gathered}$$\end{document} Floor stress components induced by the load q5:25\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} {\sigma _{x5}}= - \frac{2}{\pi }\int_{{\frac{L}{2}}}^{l} {\frac{{{q_3}{y^3}d\xi }}{{{{[{{(x - \xi )}^2}+{y^2}]}^2}}}} \hfill \\ =\frac{{{q_0}}}{{\pi e}}\left[ \begin{gathered} \frac{{{{(2x - L)}^2}y+{y^3}}}{{{{(2x - L)}^2}+4{y^2}}} - \frac{{{y^3}+y(x - l)(x - 0.5L)}}{{{{(x - l)}^2}+{y^2}}} \hfill \\ +(x - 0.5L)\left( {\arctan \frac{{2x - L}}{{2y}} - \arctan \frac{{x - l}}{y}} \right) \hfill \\ \end{gathered} \right] \hfill \\ \end{gathered}$$\end{document}26\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} {\sigma _{y5}}= - \frac{2}{\pi }\int_{{\frac{L}{2}}}^{l} {\frac{{{{(x - \xi )}^2}{q_3}y{\mathrm{d}}\xi }}{{{{[{{(x - \xi )}^2}+{y^2}]}^2}}}} \hfill \\ = - \frac{{{q_0}}}{{\pi e}}\left\{ \begin{gathered} (x - l)\left( {\arctan \frac{{2x - L}}{{2y}} - \arctan \frac{{x - l}}{y}} \right) - \frac{{{{(2x - L)}^2}y}}{{{{(2x - L)}^2}+{y^2}}}+\frac{{(x - l)(x - 0.5L)y}}{{{{(x - l)}^2}+{y^2}}} \hfill \\ +y\left[ {\ln \left| {{{(x - l)}^2}+{y^2}} \right| - \ln \left| {{{\left( {x - \frac{L}{2}} \right)}^2}+{y^2}} \right|} \right] - \frac{{{y^3}}}{{{{(x - l)}^2}+{y^2}}}+\frac{{4{y^3}}}{{{{(2x - L)}^2}+4{y^2}}} \hfill \\ \end{gathered} \right\} \hfill \\ \end{gathered}$$\end{document}27\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} {\tau _{x{y_5}}}= - \frac{2}{\pi }\int_{{\frac{L}{2}}}^{l} {\frac{{{q_2}(x - \xi ){y^2}{\mathrm{d}}\xi }}{{{{[{{(x - \xi )}^2}+{y^2}]}^2}}}} \hfill \\ =\frac{{{q_0}}}{{\pi e}}\left[ {y\left( {\arctan \frac{{2x - L}}{{2y}} - \arctan \frac{{x - l}}{y}} \right) - \frac{{(4x - 2L - 1){y^2}}}{{{{(2x - L)}^2}+4{y^2}}}+\frac{{(x - l - 1){y^2}}}{{{{(x - l)}^2}+{y^2}}}} \right] \hfill \\ \end{gathered}$$\end{document} By stress superposition, the stress at any point within the floor strata is given by:28\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left\{ {\begin{array}{*{20}{l}} {{\sigma _x}={\sigma _{{x_1}}}+{\sigma _{{x_2}}}+{\sigma _{{x_3}}}+{\sigma _{{x_4}}}+{\sigma _{{x_5}}}+{\sigma _{{x_6}}}} \\ {{\sigma _y}={\sigma _{{y_1}}}+{\sigma _{{y_2}}}+{\sigma _{{y_3}}}+{\sigma _{{y_4}}}+{\sigma _{{y_5}}}+{\sigma _{{y_6}}}} \\ {{\tau _{xy}}={\tau _{x{y_1}}}+{\tau _{x{y_2}}}+{\tau _{x{y_3}}}+{\tau _{x{y_4}}}+{\tau _{x{y_5}}}+{\tau _{x{y_6}}}} \end{array}} \right.$$\end{document} Stress distribution characteristics of the floor beneath the remnant coal pillar in the no. 11 coal seam When calculating the floor stress field, the stress induced by the self-weight of the strata should be taken into account. The modified vertical stress expression σy′:29\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\sigma _y}^\prime ={\sigma _y}+{\gamma _3}y$$\end{document} Equation (21) and Eq. (22) are solved numerically to obtain the stress distribution characteristics of the floor strata within the influence zone of the remnant coal pillar, as shown in Fig. 5. Fig. 5Stress distribution patterns in the floor stratum under remnant coal pillars of Coal Seam 11 (a) Vertical Stress Distribution Curve (b) Horizontal Stress Distribution Curve (c) Shear Stress Distribution Curve. Stress distribution patterns in the floor stratum under remnant coal pillars of Coal Seam 11 (a) Vertical Stress Distribution Curve (b) Horizontal Stress Distribution Curve (c) Shear Stress Distribution Curve. According to the numerical results in Fig. 5, the floor stress distribution within the coal-pillar influence zone exhibits the following typical characteristics: 1) The vertical stress in the floor directly beneath the coal pillar reaches its maximum at the center; as the horizontal distance increases, the stress beneath the pillar shows a trend of a sharp decrease followed by a gradual rebound, with a minimum occurring at a certain location. The distance from the minimum-stress location to the pillar center increases with increasing depth into the floor, and the vertical stress gradually approaches the in-situ stress state away from the coal pillar. 2) As the floor depth increases, the distribution of horizontal stress changes markedly: the area directly beneath the coal pillar is initially in a high-stress state, but gradually evolves into a low-stress zone, and the stress peaks shift from the central region toward both sides. When the depth exceeds 20 m, the horizontal-stress distribution tends to stabilize and no longer shows significant variation with further increases in depth. 3) In the floor strata near the coal-pillar edges, the shear stress reaches its maximum; its magnitude decays with increasing depth, and the peak location gradually spreads toward both sides. Taking the coal-pillar centerline as the boundary, compressive stress and tensile stress are observed on the two sides, respectively. From a vertical-section perspective, in the area directly beneath the coal pillar, the shear stress increases rapidly with increasing horizontal distance. However, on both sides of the coal pillar, this stress decreases gradually with increasing horizontal distance, and the rate of decay progressively slows. 4) Analysis of the stress distribution in the floor strata indicates that vertical stress is the dominant component, whereas the effects of horizontal stress and shear stress are relatively weaker, with the horizontal-stress influence zone being comparatively limited. The curves show that the peak magnitudes of all three stress components gradually decay with increasing depth and increasing distance from the coal pillar; the decay rate decreases progressively, and the stresses ultimately approach the in-situ stress state. Development and calibration of the numerical model Numerical model development Based on the actual geological and construction conditions on site, when the drivage face of the return-air roadway of the 13,402 working face in Sunjiagou Coal Mine was 1300 m away from the West Wing return-air main roadway of the No. 13 coal seam, roof borehole coring was conducted at a location 50 m from the drivage face; the coring site was 1250 m from the West Wing return-air main roadway. The recovered coal–rock cores were processed into test specimens, and rock-mechanics tests were then performed using a rock-mechanics testing machine. Fig. 6Experimental setup for mechanical-parameter testing. Experimental setup for mechanical-parameter testing. With this experiment, the mechanical properties of the rock can be determined accurately, covering its behavior under different stress conditions; meanwhile, the full geometric characteristics of fractures and cracks can be obtained, such as crack length, width, and orientation; in addition, some mechanical characteristics of fractures can be measured, for example, parameters such as fracture toughness. Analysis of the experimental data indicates that the mean compressive strength of the coal specimens is 18.7 MPa and the mean tensile strength is 1.9 MPa; according to the engineering rock-mass classification standard, this coal seam falls into the category of medium hardness. Meanwhile, the compressive strength of the roof strata reaches 103.77 MPa and the tensile strength is 10.19 MPa, and its mechanical properties meet the criteria for hard rock strata. Testing of the mechanical parameters of the coal–rock mass can provide basic mechanical data for selecting roadway support parameters. The experimental images of the mechanical-parameter tests are shown in Fig. 6: Simulation scheme and model construction This study takes the 13,402 return-air roadway of the Sunjiagou Coal Mine in Shanxi Province as the research object, and a three-dimensional numerical simulation model is established based on the actual geological conditions of the mining area. The geometric parameters and boundary conditions of the model are shown in Fig. 7. In the model, the X-direction corresponds to the strike of the working face with a length of 500 m; the Y-direction represents the dip direction with a length of 250 m; and the Z-direction corresponds to the overburden direction with a height of 82 m, within which the No. 13 coal seam is 13.4 m thick and is classified as an ultra-thick seam. To effectively reduce boundary effects in the numerical simulation, protective coal pillars were arranged around the panel in the model. To improve simulation accuracy, local mesh refinement was applied to the roadway and the surrounding-rock region. In terms of boundary conditions, a stress boundary was applied on the top surface, where a vertical stress of 7.5 MPa was imposed to represent the in-situ burial depth; displacement constraints were applied on the lateral boundaries and the bottom boundary. Based on the in-situ stress measurements for the return-air roadway, the horizontal stress was applied as 1.3 times the vertical stress. The Mohr–Coulomb strength criterion was adopted to evaluate rock-mass failure, and the mechanical parameters are listed in Table 3. In the numerical model, the No. 13 coal seam was treated as having a thickness of 13.4 m. In the engineering background, the originally exposed total thickness of the No. 13 coal seam in the 13,402 working face is 14.06 m, including a 0.7 m dirt band. To facilitate model simplification and parameter assignment, the dirt band was equivalently incorporated into the adjacent surrounding-rock strata during numerical simulation. Therefore, an effective coal seam thickness of 13.4 m was adopted. Fig. 7FLAC3D numerical model schematic diagram. FLAC3D numerical model schematic diagram. Table 3Rock mass mechanical parameters of the model.LithologyBulk modulus / GPaShear modulus / GPaDensity / Kg·m-3C / MPaΦ / (°)T / MPaLimestone10.948.926004.60366.83Mudstone9.116.8324005.03323.91Coal seam 111.800.8314001.5261.5Gray mudstone2.031.0423003.20333.04Black mudstone2.691.5323003.40333.24Coal seam 131.800.8314001.5261.5 Table 3Rock mass mechanical parameters of the model.LithologyBulk modulus / GPaShear modulus / GPaDensity / Kg·m-3C / MPaΦ / (°)T / MPaLimestone10.948.926004.60366.83Mudstone9.116.8324005.03323.91Coal seam 111.800.8314001.5261.5Gray mudstone2.031.0423003.20333.04Black mudstone2.691.5323003.40333.24Coal seam 131.800.8314001.5261.5 Rock mass mechanical parameters of the model. The simulation was carried out in stages following the actual mining sequence. First, retreat mining of the two upper panels (11202 and 11203) was simulated to represent extraction of the overlying coal seam. On this basis, the drivage of the 13,402 return-air roadway under mining disturbance was further simulated, with emphasis on the influence of the goaf formed by upper-seam extraction on roadway stability during excavation. During panel extraction, the roof was managed using the free-caving method; over time, the caved coal and rock were gradually compacted to form a load-bearing body that provides partial support to the overburden, thereby directly affecting the stress distribution around the goaf. Therefore, a double-yield constitutive model was adopted for the goaf to simulate the highly compacted state of caved gangue under free roof caving. In this study, the primary role of the double-yield model is to characterize the mechanical behavior of caved gangue in the goaf during its transition from a loose accumulation state to a gradually compacted load-bearing body. Since the compaction degree of the goaf directly affects stress transfer around the overlying remnant coal pillar and the stress redistribution characteristics of the surrounding rock in the underlying roadway, this model is used not only to describe the compaction behavior of goaf materials, but also to provide an important basis for the subsequent analysis of the effects of roadway offset distance, coal pillar width, and interburden thickness. Calibration of the double-yield model for the goaf The Double-Yield constitutive model implemented in FLAC3D is adopted in this study, as it can efficiently capture the stress–strain evolution of caved rock in the goaf during the compaction process. The Double-Yield model enables numerical simulation of the strain-hardening behavior of goaf backfill materials. According to the stress–strain relationship for the compression of broken rock masses (σ-εg) derived by Salamon11,28, the goaf material in this work is assumed to follow this relationship, which has been widely recognized and applied in previous studies. The resulting stress–strain relationship for the goaf material is presented in Table 4. Fig. 8Calibration results plot. Calibration results plot. Table 4Stress-strain relationship of gob materials.StrainStress /MPaStrainStress /MPa0.000.000.0911.480.010.760.113.940.021.600.1116.920.032.530.1220.570.043.580.1331.140.054.750.1439.200.066.090.1550.680.077.630.1668.330.089.400.1798.97 Table 4Stress-strain relationship of gob materials.StrainStress /MPaStrainStress /MPa0.000.000.0911.480.010.760.113.940.021.600.1116.920.032.530.1220.570.043.580.1331.140.054.750.1439.200.066.090.1550.680.077.630.1668.330.089.400.1798.97 Stress-strain relationship of gob materials. Following the inversion procedure for calibrating the double-yield model for goaf materials reported in Ref29., and in combination with the site conditions of the Sunjiagou Coal Mine, the calibration results of the double-yield model are obtained, as shown in Fig. 8. The final mechanical parameters of the goaf rock mass are listed in Table 5. Table 5Gob backfill parameters.Density / Kg·m-3Bulk modulus / GPaShear modulus / GPaInternal friction angle / (°)Dilation angle / (°)162814.3710.6327 Table 5Gob backfill parameters.Density / Kg·m-3Bulk modulus / GPaShear modulus / GPaInternal friction angle / (°)Dilation angle / (°)162814.3710.6327 Gob backfill parameters. Analysis of simulation results The stress distribution curves at different stratigraphic levels beneath the coal pillar were extracted from the numerical model, as shown in Fig. 9, and compared with the theoretically calculated floor stress distribution beneath the coal pillar presented in Sect. 3.3.2. Fig. 9Stress distribution curves in the floor strata beneath residual coal pillars (a) Vertical Stress Curve (b) Horizontal stress curve (c) Shear stress curve. Stress distribution curves in the floor strata beneath residual coal pillars (a) Vertical Stress Curve (b) Horizontal stress curve (c) Shear stress curve. Figure 9 illustrates the distributions of vertical stress, horizontal stress, and shear stress in the floor beneath the coal pillar in the numerical model (panels (a)–(c)). The results indicate that: (1) the vertical stress is primarily governed by the pillar self-weight and the overlying load, reaching its maximum directly beneath the pillar and gradually decaying outward, forming an “inverted rounded-cone” diffusion pattern; this trend agrees well with the theoretical curve in Sect. 3.3.2. (2) The horizontal stress is generally symmetric on both sides of the pillar floor, with higher values near the pillar edges; however, local stress anomalies may occur due to geological structures, stratigraphic anisotropy, and lateral confinement. Compared with the theoretical results, noticeable differences mainly appear at the 5 m and 10 m positions beneath the pillar, whereas the remaining regions show good consistency. (3) The shear stress arises from the interaction between vertical and horizontal stresses, and its peak occurs at the interface between the pillar edge and the floor, where abrupt changes in stress direction may facilitate shear-slip development; the shear stress then decreases outward in an annular manner, and the overall distribution is largely consistent with the theoretical analysis. Combined with Figs. 5 and 9, the theoretical analysis and numerical simulation show good consistency in terms of the floor stress distribution. Both methods indicate that the vertical stress reaches its peak in the middle region directly beneath the coal pillar and gradually decreases toward both sides along the horizontal direction. With increasing depth, the horizontal stress exhibits an evolutionary trend in which the high-stress zone shifts from the central region toward both sides. The peak shear stress is mainly concentrated near the coal pillar edges and gradually weakens with increasing depth. These results indicate that the analytical model established based on the limit equilibrium zone theory and the elastic half-plane theory can effectively reveal the basic characteristics of floor stress redistribution beneath the remnant coal pillar, while the numerical simulation further verifies this stress distribution pattern. However, certain local differences still exist between the theoretical and numerical results. This is mainly because the theoretical model treats the floor as a homogeneous and continuous elastic half-plane and adopts an idealized symmetric loading condition, whereas the numerical model further considers practical engineering factors, including the stratified structure of coal and rock strata, the compaction and load-bearing effect of the goaf, boundary constraints, and mining-induced disturbance. Therefore, some deviations in stress distribution may occur in shallow local regions or at specific depths. Overall, these differences do not affect the consistency between the two methods in terms of the general stress distribution pattern. Instead, they indicate that the numerical simulation can more realistically reflect the stress response characteristics under complex engineering conditions. Deformation and failure characteristics of roadway surrounding rock beneath remnant coal pillars To reveal the dominant mechanisms governing the surrounding-rock response of the mining roadway affected by the remnant coal pillar, a parametric analysis was conducted around three key factors: roadway offset distance, remnant coal pillar width, and interburden thickness. These three factors have different engineering implications. The roadway offset distance directly determines whether the roadway enters the high-stress influence zone of the remnant coal pillar, and is therefore the most sensitive control parameter during roadway layout design. The remnant coal pillar width reflects the integrity of the overlying load-bearing structure and the degree of stress concentration, serving as an important structural factor controlling the intensity of additional loading. The interburden thickness characterizes the geometric condition for downward disturbance propagation and is a key geological parameter affecting the disturbed depth and range of the underlying roadway. On this basis, the evolution characteristics of the surrounding-rock plastic zone, stress distribution, and failure depth under variations in these three factors were investigated, providing targeted guidance for roadway layout optimization and subsequent adjustment of support parameters. Deformation and failure characteristics of the retreat roadway under different offset distances To facilitate the investigation of deformation and failure characteristics of the retreat roadway under different offset distances, as well as the numerical computations, the following layout schemes are proposed. Scheme 1 adopts an inward offset of 15 m, located 15 m to the left of the coal pillar; Scheme 2 adopts an inward offset of 10 m, located 10 m to the left of the coal pillar; Scheme 3 adopts an inward offset of 5 m, located 5 m to the left of the coal pillar; Scheme 4 adopts an inward offset of 0 m, i.e., the roadway is arranged to overlap with the roadway in the No. 11 coal seam; Scheme 5 adopts an outward offset of 5 m, i.e., beneath the left side of the coal pillar; Scheme 6 adopts an outward offset of 10 m, i.e., directly beneath the coal pillar. Schemes 7–11 are arranged on the right side of the coal pillar and are symmetrically distributed with respect to Scheme 6 as the centerline. Figure 10 shows the layout schemes of the retreat roadway. Fig. 10Gate roadway layout plan for No. 13 coal seam. Gate roadway layout plan for No. 13 coal seam. The plastic-zone patterns of the roadway surrounding rock vary markedly with offset distance, indicating that selecting an appropriate roadway position can effectively reduce the degree of surrounding-rock damage. Accordingly, the plastic-zone distribution around the 13,402 roadway under different offset distances is analyzed, as shown in Fig. 10. As illustrated in Fig. 11, the plastic zones of the roadway surrounding rock exhibit mirror symmetry about the centerline of Scheme 6. For Schemes 1–5, the plastic zone expands from the lower part of the left rib and the upper part of the roof toward the upper part of the left rib and the upper-left roof corner; that is, the plastic-zone distribution shows a counterclockwise rotation about the roadway diagonal. To further verify this rotation pattern, the stress distribution in the roadway surrounding rock under different offset distances is investigated in conjunction with the three stress components discussed in Chap. 3. Following the procedure used for Fig. 11, the vertical stress in the roadway surrounding rock under different offset distances is extracted, as shown in Fig. 12. Figure 12 shows a comparative distribution of vertical stress in the surrounding rock of the retreat roadway under different offset distances. The vertical-stress pattern is similar to that of the plastic-zone distribution. At the location of Scheme 6, the vertical stress is symmetrically distributed about the roadway centerline; moreover, taking the centerline of Scheme 6 as the axis, the vertical-stress distributions of the roadways on both sides exhibit mirror symmetry. For the roadway schemes on the left side, compressive-stress concentration occurs on the right side of the roadway, and the concentration becomes more pronounced as the offset distance decreases, with the corresponding stress magnitude increasing. Similarly, for the roadway schemes on the right side, compressive-stress concentration occurs on the left side of the roadway, and the stress magnitude also increases as the offset distance decreases. Fig. 11Comparative distribution maps of plastic zones in gate roadway surrounding rock under varied dislocation distances. Comparative distribution maps of plastic zones in gate roadway surrounding rock under varied dislocation distances. Fig. 12Comparative vertical stress distribution maps in gate roadway surrounding rock under varied dislocation distances. Comparative vertical stress distribution maps in gate roadway surrounding rock under varied dislocation distances. By comparing the vertical-stress distributions under different offset distances, it is found that the stresses on both sides are mirror-symmetric about the roadway centerline directly beneath the coal pillar, consistent with the plastic-zone distribution. The vertical stress and shear stress increase as the offset distance decreases, showing a negative correlation with the offset distance. Based on the above analysis, a rational selection of roadway location under different offset distances can effectively reduce the extent of the roadway plastic zone. By establishing 11 numerical models with different offset distances, the deformation and failure characteristics of the retreat roadway under varying offsets are obtained. The plastic zone is mirror-symmetric about the centerline of the roadway directly beneath the coal pillar (Scheme 6). When an inward offset is arranged on the left side of the coal pillar, a smaller offset distance leads to a more pronounced expansion of the plastic zone (e.g., for a 15 m inward offset, the roof failure depth reaches 3.5 m). When an outward offset is adopted (beneath the coal pillar), the plastic-zone extent is relatively small, and the roadway located at the middle beneath the coal pillar (Scheme 6) shows the best stability, with a roof failure depth of only 2 m. The vertical stress and shear stress increase as the offset distance decreases, whereas the horizontal stress decreases. The stress distribution agrees with the plastic-zone morphology: the maximum principal-stress concentration zone overlaps with the plastic failure zone. In selecting the roadway position, priority should be given to a small outward offset or a central layout beneath the coal pillar, whereas inward-offset layouts require strengthened support. Under the conditions of the engineering model established for the 13,402 return airway in Sunjiagou Coal Mine, an outward-offset layout or a layout directly beneath the coal pillar can effectively reduce the extent of the surrounding-rock plastic zone, with the layout directly beneath the coal pillar exhibiting better stability. This conclusion is applicable to similar engineering conditions involving closely spaced coal seam mining, pronounced influence of overlying remnant coal pillars, and asymmetric stress distribution in the surrounding rock. For roadways with different burial depths, coal–rock structures, and support conditions, the optimal layout position should still be further analyzed based on specific engineering parameters. Deformation and failure characteristics of the retreat roadway under different remnant coal-pillar sizes To facilitate the investigation of deformation and failure characteristics of the retreat roadway under different coal-pillar sizes, as well as the numerical computations, the following schemes are proposed: Scheme 1 adopts a coal-pillar width of 5 m with a 5 m inward roadway offset; Scheme 2 adopts a coal-pillar width of 10 m with a 5 m inward roadway offset; Scheme 3 adopts a coal-pillar width of 15 m with a 5 m inward roadway offset; Scheme 4 adopts a coal-pillar width of 20 m with a 5 m inward roadway offset; Scheme 5 adopts a coal-pillar width of 25 m with a 5 m inward roadway offset; and Scheme 6 adopts a coal-pillar width of 30 m with a 5 m inward roadway offset. The layout schemes of the retreat roadway are shown in Fig. 13. A numerical simulation was conducted to investigate the deformation and failure characteristics of the retreat roadway under different coal-pillar sizes. Six numerical models with different pillar sizes were established, and two processes were simulated in each model: (i) retreat mining of the No. 11 coal seam to form a remnant coal pillar, and (ii) excavation of the 13,402 return-air roadway. After the simulations were completed, post-processing was performed to extract the plastic-zone contour maps of the remnant pillar and the 13,402 return-air roadway. The plotted domain was taken as a section at y = 125 m; x within 210–295 m; and z within 15–70 m. Fig. 13Gate roadway layout plan diagrams under varied coal pillar sizes. Gate roadway layout plan diagrams under varied coal pillar sizes. Overall, the plastic-zone patterns of the roadway surrounding rock differ markedly with coal-pillar size, indicating that, in multi-seam mining, different engineering layouts in the upper seam can impose different levels of influence on the roadways and working faces of the lower seam. A comparative analysis of the plastic-zone distributions around the 13,402 roadway under different coal-pillar sizes is presented in Fig. 14. As shown in Fig. 14, with increasing coal-pillar size, the plastic zones in the roadway roof and ribs progressively expand. When the pillar size reaches 25 m, the expansion of the roadway plastic zone slows, and the plastic-zone extent in the two ribs begins to decrease. When the pillar size further increases to 30 m, the plastic-zone extent in the ribs continues to shrink. Among the six simulation cases, the roadway plastic zone is smallest for the 5 m pillar, followed by the 30 m pillar. Fig. 14Comparative distribution maps of plastic zones in gate roadway surrounding rock under varied coal pillar sizes. Comparative distribution maps of plastic zones in gate roadway surrounding rock under varied coal pillar sizes. Based on the six numerical models with different remnant coal-pillar sizes, the following findings are obtained. For small pillars (≤ 15 m), the plastic zone penetrates through the pillar, indicating poor pillar integrity, and the roof failure depth increases with pillar size (2.5 m for a 5 m pillar and 3.5 m for a 15 m pillar). For the intermediate pillar size (20 m), damage is pronounced on both sides, while the middle part does not fully penetrate; meanwhile, the plastic zone at the roadway right rib becomes enlarged. For large pillars (≥ 25 m), pillar integrity improves and plastic-zone expansion is restrained; for the 30 m pillar, the rib failure depth is only about 1–2 m. From an engineering perspective, a coal-pillar size of 25–30 m is more favorable for maintaining the stability of the retreat roadway in the lower seam. From the simulation results obtained under the present engineering conditions, a coal-pillar width of 25–30 m is more favorable for maintaining the stability of the retreat roadway in the lower seam. It should be emphasized that this recommended range is derived under the condition that the roadway is arranged at an inward offset of 5 m on the left side of the coal pillar, and is therefore mainly applicable to engineering scenarios with similar seam occurrence, remnant-pillar influence, and roadway layout conditions. Deformation and failure characteristics of the retreat roadway under different seam spacings To facilitate the investigation of deformation and failure of the retreat roadway under different seam spacings, as well as the associated numerical computations, five simulation schemes were designed. In Scheme 1, the seam spacing is 10 m and the roadway is arranged with an inward offset of 5 m; in Scheme 2, the seam spacing is 20 m with an inward offset of 5 m; in Scheme 3, the seam spacing is 25 m with an inward offset of 5 m; in Scheme 4, the seam spacing is 30 m with an inward offset of 5 m; and in Scheme 5, the seam spacing is 35 m with an inward offset of 5 m. The roadway layout schemes are shown in Fig. 15. Fig. 15Gate roadway layout plan diagrams under varied seam spacings. Gate roadway layout plan diagrams under varied seam spacings. The plastic-zone distributions of the roadway surrounding rock differ markedly with seam spacing, and the influence of the remnant coal pillar on the lower seam gradually weakens as the seam spacing increases. A comparative analysis of the plastic-zone distribution around the 13,402 roadway under different seam spacings is presented in Fig. 16. As shown in Fig. 16, increasing seam spacing leads to a progressive reduction in the plastic-zone extent in the roadway roof and ribs, indicating a diminishing influence of the remnant pillar on the lower seam. Among the considered seam spacings, the 30 m case exhibits the smallest plastic-zone extent. Fig. 16Comparative distribution maps of plastic zones in gate roadway surrounding rock under varied seam spacings. Comparative distribution maps of plastic zones in gate roadway surrounding rock under varied seam spacings. Based on the five numerical models with different seam spacings, the main conclusions are as follows. When the seam spacing is small (e.g., 10 m), the plastic zone of the roadway roof becomes connected with the plastic zone of the upper-seam goaf, and the failure depth can reach 10 m, implying a high risk level. When the seam spacing increases to 35 m, the roof failure depth decreases to below 2.5 m, while the rib failure depth is about 1–2 m, demonstrating a pronounced improvement in stability. When the seam spacing is ≥ 30 m, the disturbance imposed by the remnant coal pillar on the roadway in the lower seam can be effectively reduced. Based on the numerical simulation results under different roadway offset distances, coal pillar widths, and interburden thicknesses, the surrounding-rock response of the 13,402 return airway exhibits relatively clear variation patterns controlled by these three factors. For the roadway offset distance, when the roadway is arranged from the outside of the coal pillar toward the coal pillar influence zone with an inward offset, the plastic zones in the roof and ribs gradually expand, and the vertical stress and shear stress continuously increase. In contrast, the surrounding rock shows relatively better stability when the roadway is located directly beneath the coal pillar or arranged with a small outward offset. For the coal pillar width, under small-pillar conditions, the integrity of the coal pillar is poor and the plastic zone is prone to penetration. As the coal pillar width increases, surrounding-rock failure first intensifies and then weakens. When the coal pillar width reaches 25–30 m, the integrity of the coal pillar and the stability of the underlying roadway become relatively well coordinated. For the interburden thickness, as the interburden thickness increases, the disturbance of the overlying remnant coal pillar on the underlying roadway gradually weakens. When the interburden thickness is no less than 30 m, the plastic zones in the roof and both ribs are significantly reduced. Overall, although the degree of surrounding-rock failure varies under different working conditions, failure is mainly concentrated in the roof and the right rib, providing a unified basis for strengthening the roof and right rib in the subsequent support optimization. Determination of optimized support parameters for the trial roadway Identification of the reinforcement support zone For the return airway of the 13,402 working face, support-parameter optimization was carried out on the basis of the original support scheme. At present, the roadway is arranged at an inward offset of 5 m on the left side of the coal pillar. According to the borehole observation results presented above, roof fractures are mainly developed within the range of 0–5 m. Meanwhile, the numerical simulation results indicate that, under different influencing factors, plastic failure of the surrounding rock is mainly concentrated in the roof and the right rib.In the original support scheme, the short roof cables are 5200 mm long, serving to penetrate the shallow fractured zone and provide effective confinement for the surrounding rock within this range. The long roof cables, with a length of 8200 mm, are mainly designed to anchor further into the deeper, relatively intact and stable surrounding rock, thereby forming a coordinated load-bearing structure composed of long and short cables. Therefore, the support optimization does not simply involve changing the cable length. Instead, while maintaining the coordinated support form of long and short roof cables, the optimization focuses on adjusting the cable layout density, spacing, and rib reinforcement parameters, so as to enhance the control effect on the shallow fractured zone and weak parts of the surrounding rock.Based on engineering background identification, field structural detection, and numerical simulation analysis, targeted optimization of the roof and right-rib support parameters of the 13,402 return airway was carried out in this study. Figure 17 presents the plastic-zone distribution characteristics of the roadway surrounding rock under different factors. Panel (a) shows the plastic-zone distribution and the stress distribution at the actual roadway location, indicating that the plastic failure is mainly concentrated in the roadway roof and the right rib; the five stress-concentration regions are also mainly distributed in the roadway roof and the lower part of the right rib. Panel (b) shows the plastic-zone distribution under different coal-pillar sizes and seam spacings, from which it can be observed that, when the roadway is arranged at an inward offset distance of 5 m (left side of the coal pillar), the plastic zone remains concentrated in the roof and the right rib under various conditions. Panel (c) shows the plastic-zone distribution at different advance roadway positions, indicating that the plastic-zone extent decreases as the advance distance increases; however, the damage is still mainly concentrated in the roadway roof and the right rib. Therefore, considering the actual conditions and the plastic-zone distribution characteristics under different factors, reinforcement support should be strengthened for the roadway roof and the right rib. Under the actual layout conditions of the 13,402 return airway, the roadway is arranged at an inward offset on the left side of the remnant coal pillar, and the surrounding rock exhibits pronounced asymmetric stress characteristics. The additional stress induced by the remnant coal pillar mainly acts on the right side of the roadway, resulting in more evident stress concentration and plastic failure in the right rib and rib-corner areas. Meanwhile, under the combined effects of stress transfer from the remnant coal pillar and mining-induced disturbance, the roof also develops relatively extensive fractures and plastic failure. Therefore, the support optimization primarily focuses on strengthening the roof and the right rib. Fig. 17Distribution law of the plastic zone in roadway surrounding rock under different factors. (a) Plastic zone and stress concentrations (b) Plastic zones for varying pillar sizes and seam spacings (c) Plastic zones at different advance positions. Distribution law of the plastic zone in roadway surrounding rock under different factors. (a) Plastic zone and stress concentrations (b) Plastic zones for varying pillar sizes and seam spacings (c) Plastic zones at different advance positions. Based on the plastic-zone damage obtained from the numerical simulations and the theoretically calculated support parameters, the following optimized support design is proposed for the 13,402 return-air roadway at Sunjiagou Coal Mine. (1) Roadway roof. Short cable bolts: Five Ф21.8 × 5200 mm high-strength, low-relaxation prestressed steel-strand cable bolts (1 × 19 construction) are installed in the roof, with a spacing of 1000 mm and a row spacing of 1000 mm. Long cable bolts: Two Ф21.8 × 8200 mm high-strength, low-relaxation prestressed steel-strand cable bolts (1 × 19 construction) are installed in the roof, with a spacing of 2000 mm and a row spacing of 1000 mm. (2) Roadway ribs. Coal-pillar rib: Four Ф22 × 2400 mm ribless threaded-steel resin bolts are installed, with a spacing/row spacing of 900 mm × 1000 mm; additionally, two Ф21.8 × 4200 mm high-strength, low-relaxation prestressed steel-strand cable bolts (1 × 19 construction) are used for reinforcement, with a spacing of 1800 mm and a row spacing of 1000 mm. Mining rib: Four Ф22 × 2400 mm ribless threaded-steel resin bolts are installed, with a spacing/row spacing of 900 mm × 1000 mm. The optimized support layout is shown in Fig. 18. Compared with the original support scheme, the roof short cable bolts are reduced from six to five, and the spacing of the roof long cable bolts is decreased from 2200 mm to 2000 mm. For the ribs, the bolt spacing is reduced from 1000 mm to 900 mm; the spacing of the reinforced long cable bolts on the right rib is reduced from 2000 mm to 1800 mm, and their overall arrangement is shifted downward toward the lower part of the roadway. Among them, the reinforcement intensity of the right-rib support parameters is greater than that of the left rib. This is mainly because the numerical simulation results show that the lower part of the right rib and the rib-corner area are the main zones of plastic failure and stress concentration, representing asymmetric weak zones under the influence of the remnant coal pillar. In contrast, the failure degree of the left rib is relatively slight. Therefore, a higher-density arrangement of reinforced cables is adopted for the right rib, and the support position is shifted downward to enhance the control effect on the weak zone in the lower part of the rib. Fig. 18Schematic diagram of parameter optimization layout for roadway support. (a) Cross-sectional view of the optimized roadway support layout (b) Plan view of the optimized roadway support layout. Schematic diagram of parameter optimization layout for roadway support. (a) Cross-sectional view of the optimized roadway support layout (b) Plan view of the optimized roadway support layout. Monitoring of surrounding-rock control performance To verify the field applicability and control effect of the optimized support-parameter scheme for the return airway of the 13,402 working face, a 20 m-long roadway section within the range of 50–70 m from the excavation face was selected as the field test section. In this section, surrounding-rock convergence monitoring, roof separation monitoring, and cable anchorage-force testing were conducted, as shown in Fig. 19. The test section is located in an area typically affected by the remnant coal pillar, and its surrounding-rock lithological assemblage, roadway cross-section, and support conditions are generally consistent with the research object of this study, indicating good engineering representativeness. The surrounding-rock control problems under the original support conditions were mainly identified based on field engineering phenomena, borehole observation results, and numerical simulation analysis. On this basis, the optimized support scheme was proposed and subsequently verified through field monitoring. Fig. 19Industrial field testing site (a) Roadway excavation face support site diagram (b) Anchor cable anchoring force test. Industrial field testing site (a) Roadway excavation face support site diagram (b) Anchor cable anchoring force test. An industrial field trial was carried out in the return-air roadway of the 13,402 working face. The convergence deformation of the roadway surrounding rock was monitored, and the results are shown in Fig. 20. As indicated by Fig. 20 (a–c), the convergence deformation of the roadway surrounding rock can be divided into three stages: a rapid-deformation stage, a decelerating-deformation stage, and a stable-deformation stage. During 0–20 d of monitoring, the deformation increased sharply, with roof subsidence rising to 2 mm, floor heave to 1 mm, and the convergence of the left and right ribs to approximately 1.5 mm. During 20–60 d, the deformation gradually slowed, indicating that the support system could effectively mobilize its load-bearing capacity. After 60 d, the surrounding-rock deformation reached its peak and became essentially stable; the final deformations of the roof–floor and the two ribs were all within 4–5 mm, demonstrating that the proposed optimized support scheme can ensure the stability of the roadway surrounding rock. Fig. 20Results of industrial field testing (a) Surrounding rock deformation convergence No.1 measuring point (b) Surrounding rock deformation convergence No.2 measuring point (c) Surrounding rock deformation convergence No.3 measuring point. Results of industrial field testing (a) Surrounding rock deformation convergence No.1 measuring point (b) Surrounding rock deformation convergence No.2 measuring point (c) Surrounding rock deformation convergence No.3 measuring point. Conclusions Targeting stability control of retreat roadways beneath remnant coal pillars in closely spaced coal seams, the 13,402 return-air roadway of Sunjiagou Coal Mine (Shanxi Province, China) was investigated through an integrated program combining theoretical analysis, three-dimensional numerical simulation, in-situ observation, and industrial-scale validation. The principal conclusions are summarised as follows. (1) On the basis of limit-equilibrium zone theory and elastic half-plane theory, a coal-pillar load-bearing model and analytical solutions for floor stress were derived, clarifying the three-dimensional stress characteristics induced by the remnant pillar. The floor directly beneath the pillar forms a vertical-stress concentration zone, with a peak value of 24.1 MPa. Shear-stress maxima develop near the pillar edges due to abrupt changes in stress direction, and the maximum shear stress attenuates with increasing depth. With depth, horizontal stress redistributes from the pillar centre toward both sides and becomes essentially stable below 20 m, providing a theoretical basis for roadway positioning. (2) Three-dimensional FLAC3D simulations indicate that roadway offset distance, coal pillar width, and seam spacing all significantly affect the stability of the retreat roadway in the lower seam. Under the engineering conditions considered in this study, an outward-offset layout or a layout directly beneath the coal pillar is more favorable for controlling plastic-zone expansion. When the coal pillar width is 25–30 m and the seam spacing is not less than 30 m, the stability of the retreat roadway in the lower seam is relatively better. These conclusions are applicable to engineering scenarios with similar closely spaced seams, significant remnant-pillar influence, and obvious asymmetric loading on the surrounding rock. (3) Borehole endoscopic observations showed that roof fractures in the 13,402 return-air roadway were mainly concentrated within 0–5 m, implying relatively intact shallow strata but necessitating targeted reinforcement. By integrating numerical predictions with site observations, an optimised support scheme featuring coordinated long–short roof cable bolts and reinforced rib bolting was proposed. The roof short cable bolts were adjusted to five (spacing 1000 mm), the roof long cable-bolt spacing was set to 2000 mm, rib bolt spacing was reduced to 900 mm, and the reinforcement cable-bolt spacing on the coal-pillar rib was reduced to 1800 mm, achieving improved compatibility between support parameters and the local stress environment. (4) Industrial trial monitoring confirmed the effectiveness of the optimised support. Deformations of the roof, floor, and both ribs were controlled within 4–5 mm over the monitoring period. Roof separation remained zero at all three deep-point monitoring stations, cable-bolt anchorage force was stable, and no support failure was observed, satisfying the requirements for safe and efficient roadway service. Overall, the study elucidates the coupled mechanism of stress transfer, plastic-zone development, and deformation–failure evolution for roadways beneath remnant pillars in closely spaced seams, and delivers a field-deployable support optimisation strategy with both mechanistic grounding and demonstrated engineering effectiveness.