The shield-roof interaction as mining proceeds in longwall panels remains unclear, hindering the further increase of longwall productivity. To uncover the mechanisms of shield-roof interaction, using our self-developed Status of Shield and Roof IntelliSense (SSRI) system, we investigated the effects of idle time, retreating rate, setting pressure, yielding, and shearer’s cutting, as well as neighboring shields’ advance on the spatial-temporal evolution of leg pressure and leg closure of shields. Our results show that the shield-roof interaction is not only dependent on the shield capacity, but also collectively determined by the time-related factors, the geological condition, the setting pressure, yielding characteristics, and mining method. Understanding the shield-roof interaction in longwall panels enables us to apply the SSRI system for ground control in longwall coal mines. Early warning of severe roof weighting can be achieved by establishing a warning model based on the decision tree algorithm. Apart from this, we can also assess the working condition of yield valve and diagnose fluid leakage of shield cylinder using the SSRI system. Finally, we propose the research prospects on shield-roof interaction in longwall panels to achieve a more reasonable determination of shield capacity, prediction of roof fall and coal wall spalling, and self-adaptive control of the shield.

1. Introduction

Longwall mining is currently the predominant coal mining method in China, thanks to its high efficiency of coal production [1, 2]. The recent advances in large-cutting-height mining has launched a new era of longwall coal mining [3]. For instance, Shangwan mine located in Inner Mongolia started the coal production in the Panel LW 12401 using a record-breaking large-mining-height of 8.8 m on 19 March 2018. Large-cutting-height mining requires strong shield to support the extraordinarily high roof loading, and therefore, the shield capacity in this panel reaches 26,000 kN by increasing the diameter of shield leg. Nevertheless, a strong shield does not necessarily address all the issues in the longwall panel with large-cutting-height. Shield crushing still occurs during the retreating of some panels [4, 5]. It is also reported that the roof between the front tip of the shield canopy and the coal face falls occasionally [3, 5]. The unwanted idle time caused by roof falls and shield malfunctions hinders the additional growth of longwall productivity. The occurrence of these problems indicates our limited understanding on the shield-roof interaction.

For decades, many researchers have proposed various methods, models, and theories to interpret the shield-roof interaction and to determine the shield capacity. For example, detached roof block method [6, 7], yielding foundation model [8], design of powered support selection model [9], ground response curves [10], “Roof -support-floor” system model [11], and Voussoir beam theory [12]. These studies have deepened our understanding on the roof activities and the interplay between shield and roof. Nevertheless, these results which are based on idealized deduction can be hardly applied in practice, because of complex geological conditions underground.

The development of monitoring and computer techniques enables us to continuously record a fair amount of data of the panel production and shield resistance. A comprehensive analysis of the recorded data can help us to gain a deeper understanding on the shield-roof interaction. Many researchers have made such attempts to analyze the shield-roof interaction by developing various computer programs [1317]. It is worth mentioning that has conducted a comprehensive analysis using their self-developed longwall visual analysis (LVA) software on shield-roof interaction based on Australian longwall coal mines [1821]. Moreover, they put forward the methods for predicting roof weightings and roof falls. These studies have significantly advanced the application of monitoring and computer techniques in coal mine ground control.

However, a better interpretation of the recorded data is impeded primarily due to the following issues. First, current analysis on the tempoevolution of leg pressure in a single-shield supporting cycle does not extract enough information for characterizing the shield-roof interaction. Researchers typically interpret the time-weighted average pressure (TWAP), number of yielding events, setting pressure, and final pressure from a single shield cycle [13, 1517, 20]. These parameters are however far from enough to help us understand the complex shield-roof interaction. Second, the supporting characteristics of one shield should be correlated with the neighboring shields. The roof stratum is not a series of discrete blocks with their widths being equivalent to the shield width, as assumed in the current shield capacity calculation methods [5]. Instead, the roof stratum is more likely to be a continuous block acting on a set of shields, and all the shields resist the roof deformation collectively. Any change of the load on one shield is expected to affect the other adjacent shields. Third, existing studies only collect the total number of yielding events in each shield cycle [21], but overlook other important yielding characteristics, such as the leg pressure and the time duration of yielding events. These characteristics are useful to shed lights on the shield-loading quality and the shield-roof interaction. Fourth, both the leg pressure and leg closure should be considered to better interpret the shield-roof interaction. Yielding of the shield occurs when the roof loading exceeds the load capacity of a shield by yield valve open and leg closure. The roof displacement can be negligible if yielding does not occur, while the leg closure primarily determines the roof displacement in the case of yielding. Finally, it is believed that other factors, such as time, the geological condition, and the mining method can greatly influence the supporting characteristics of shield [2224], while few studies accounting for these factors have been conducted.

Although extensive studies have analyzed the shield-roof interaction, the mechanisms of shield-roof interaction remain unclear considering the above remaining issues. The objective of this study is to uncover the mechanisms of shield-roof interaction and improve our ability of roof control during longwall coal mining. Based on the self-developed Status of Shield and Roof IntelliSense (SSRI) system, we first investigate the characteristic parameters of shield measured in the fields considering the effects of idle time and longwall retreat rate, setting pressure, yielding, shearer’s cutting, and advance of neighboring shields. Second, we present the preliminary application of SSRI system in roof control regarding early warning of severe roof weightings and assessment of supporting quality. Finally, based on the state-of-the-art results, we make some recommendations for future study on shield-roof interaction in longwall panels.

2. Methods: Longwall Shield Monitoring and Data Analysis

Several techniques can be used to monitor the loading characteristics of shields in longwall panels. The leg pressure sensor is usually provided by the electrohydraulic shield manufacturer to continuously measure the leg pressure. The frequency of data recording, which depends on the manufacturer, varies from one data per second to several minutes. It is inferred that a finer time increment would be more helpful in investigating the shield-strata interaction because more valuable data would be provided. In addition, the infrared system is usually installed on the shield and shearer to trace the location and moving direction of the shearer. The reed rod sensor can be used to document the DA ram extension distance. Another sensor is developed in the present study to measure the real-time leg closure [25].

The shield cycle refers to the time duration which starts from the initial setting of the leg pressure and ends with the subsequent leg setting (Figure 1(a)). Therefore, during each cycle, the shield experiences first setting, loading, unloading, and forward-moving. Based on the loading/unloading characteristics of the shields, researchers have classified the shield cycle into three distinct stages: setting period, roof deformation period, and supporting area change period [22, 23]. Other researchers proposed four stages instead: initial period, relatively stable period, cutting-influenced period, and neighboring shield movement period [26]. These studies shed lights on the supporting behavior of shields. Based on the proposed stages of the shield cycle, we characterize the shield loading and displacement from the shield supporting cycle which is obtained from the SSRI system. The SSRI system can extract more than 20 characteristic parameters during a shield-supporting cycle. We define the parameters used in this paper as follows:

2.1. Setting Pressure and Revised Setting Pressure

The setting pressure is the active pressure that a shield initially provides to support the roof when the shield canopy just set against the roof and the hydraulic fluid supply ends (point ps in the insets of Figure 1(a)). After the inflow of hydraulic fluid is terminated, it is common to observe that the shield pressure decreases slightly (0.5∼5 MPa) and quickly (1∼3 mins) once the filling is terminated after the predetermined value is achieved (Figure 1(a)). The reduction of shield pressure in this period can be attributed to several reasons. On the one hand, crushing of the coal/rock pieces on the top of the shield canopy or below the shield base causes the extension of the leg cylinder thereby decreasing the leg pressure slightly since no more hydraulic fluid is supplemented. On the other hand, the extension of leg cylinder does not terminate immediately when the valve is closed due to inertia, which may also contribute to the decrease of leg pressure. We define the shield pressure at the new steady state as the revised setting pressure. Unless otherwise stated, setting pressure in the context of this paper is the revised setting pressure.

2.2. Loading Rate in Each Period

The loading rate in each period of the supporting cycle is shown in Figure 1(a) along with the evolution of shield resistance, which is calculated every five minutes. The loading rate in the initial period is the load-increasing rate shortly after initial setting of the shield (normally 5–20 mins). It depends on the intrinsic properties of shield and geological conditions, so it is a measure of the severity of roof activity. The loading rate at relatively stable period normally keeps at a low level. In most cases, this period occupies most time of the supporting cycle, and therefore, the loading rate in this period is critical for evaluating the supporting characteristics of a shield. The loading rate during cutting-influenced period and neighboring shields’ advance period describes the load increasing rate when the shearer drum approaches and passes the shield of interest and its neighboring shields advance.

2.3. Final Pressure and Time-Weighted Average Pressure (TWAP)

The final pressure of a shield (pf in Figure 1(a)) attains when its neighboring shields are lowered and advanced, which is normally considered as the maximum pressure in the whole supporting cycle. When there are few or no yielding events during the supporting cycle, the time-weighted average resistance (TWAP) considering the temporal evolution of shield resistance is more suitable for characterizing the supporting characteristics. An example of the temporal evolution of these two parameters is shown in Figure 2.

2.4. Leg Closure of the Shield

When the setting pressure is achieved, the nonreturn valve is closed and the hydraulic fluid inside the cylinder is in a closed state and supports the shield leg. In the whole supporting cycle, leg closure can occur due to unloading during yielding, the elastic compression of hydraulic fluid, and sometimes minor leakage of hydraulic fluid.

2.5. Parameters Associated with Yielding

When the roof loading on the shield reach the shield capacity, the yield valve opens to prevent the shield from damaging by expelling some of the high-pressure fluid out of the cylinder to lower down the fluid pressure in the cylinder. Afterwards, the yield valve closes. The time interval in between the open and close of the yield valve is referred to as a yielding cycle. Detailed definition of each parameter associated with yielding is as follow:(a)The pressure at which the yield valve opens (pyo in Figure 3)(b)The pressure at which the yield valve closes, that is, when a yielding cycle ends (pyc in Figure 3)(c)The pressure drop during a yielding cycle is the reduction of shield resistance in between the open and the subsequent close of the yield valve (Δpa in Figure 3)(d)The pressure increment in between two consecutive yielding events is the increase of shield resistance from the time when the yield valve closes to the time when the yield valve opens (Δpb in Figure 3)(e)The duration of a yielding cycle is the time interval starting from the open to the subsequent close of the yield vale (ta in Figure 3)(f)The load-increasing period from the end of the last yielding to the beginning of the next yielding (tb in Figure 3)

Yielding contributes mostly to the total leg closure, so it is vital to investigate the leg closure characteristics during each yield valve open circumstance. Parameters including the number of yielding, yield pressure, and average pressure drop during yielding of LW223046 in Daliuta mine during 520 h are shown in Figure 4. 4407 among 53029 working cycles experience yield valve open. Specifically, the number of yielding ranges from 1 to 18 with an average value of 3.06 (Figure 4(a)). The yield pressure is between 44.1 MPa and 60 MPa with an average value of 46.37 MPa (Figure 4(b)). The average pressure drop during yield valve open is 0.9 MPa∼7.8 MPa with an average of 2.17 MPa (Figure 4(c)). The working status of any shield in any time and the severity of roof weighting can be readily obtained by analyzing the parameters associated with yielding.

Notations: pyo-pressure at which yield valve opens, pyc-pressure at which yield valve closes, pf-final pressure, Δpa-pressure drop during a yielding cycle, Δpb-pressure increment in between two consecutive yielding events, ta-duration of a yielding cycle, and tb-load increasing period in between two consecutive yielding events.

2.6. Parameters Associated with Periodic Roof Weighting

When the roof of the longwall panel is hard and strong, the periodic caving of the strong roof leads to severe and sudden roof weightings, probably causing shield damage, roof fall, and even casualties [23]. To achieve the early warning of severe roof weighting, the following parameters are used.(a)Length of retreat in between two consecutive roof weights(b)Periodic weighting interval: time interval in between two consecutives roof weightings(c)Periodic weighting intensity factor: the ratio between TWASD in weighting and nonweighting periods(d)Supporting density: shield load divided by the area of shield canopy(e)TWASD: time-weighted average supporting density(f)Number of supporting cycles experiencing roof weighting

A typical set of results of these parameters are shown in Figure 5.

3. Sites Description

The SSRI system has been widely used in 23 longwall panels in China and the U.S. since 2014. The data presented in this study are collected from 11 longwall panels among these panels. Nine panels are situated at the Shenfu-Dongsheng coal field, and one panel is in the Yongcheng coal mine. The other two panels are at the Appalachian Coal Basin in the U.S. These panels are thought to be representative of geological conditions, mining techniques, and the equipment standards of modern longwall panels.

The main parameters of these longwall panels are listed in Table 1. Take note that all the panels are equipped with two-leg hydraulic shields. The lithological strata of these three panels are presented in Table 2. Daliuta mine has a strong roof, showing conspicuous and severe periodic roof weightings, and the yield valve opens during weightings (shield capacity is 1500 t). Panel LW 63 in Cumberland mine has a medium-weak roof, showing minor roof weightings, and the yield valve almost keeps closed (shield capacity is 900 t). Panel LW203 in Yongcheng mine has a very weak roof. The thin bedrock is weathered severely with thick overlaying loose layers. The designed shield capacity is only 400 t, which is not sufficient to support the roof. Yield valve opens frequently, and thus large leg closure occurs.

4. Results and Discussion: Factors Influencing the Supporting Characteristics of Shield

4.1. Effects of Idle Time and Longwall Retreating Rate

The time-dependent deformation and failure of roof strata can significantly influence the shield-roof interaction [23]. In longwall coal mining, the retreating process only lasts for tens of minutes to several hours, while the idle time occupies longer time due to various reasons such as regular maintenance of machinery and unpredictable decay caused by adverse underground environments. It is therefore important to consider the effect of idle time on supporting characteristics of shield. In the following, we analyze the effect of idle time under two roof conditions: strong roof and medium-weak roof. The simplified lithological strata for panels with a strong roof (Panel LW22304 in Daliuta mine) and a medium-weak roof (Panel LW63 in Cumberland mine) are shown in Table 2.

A continuous spatial-temporal evolution of shield resistance during 130 h in Panel LW22304 of Daliuta mine is shown in Figure 6(a). In between this period, there are two long idle times, lasting for 40 h and 25 h, respectively. Because of different roof-weighting characteristics, there are two distinct evolution characteristics of shield resistance along the dip direction. In the first case, the shield resistance increases rapidly after initial setting (No. 20–60 shields in Figure 6(a)) and reaches the shield capacity. During this period, the yield valve frequently opens, and thus the cumulative roof convergence is obvious. For No. 70–150 shields, after initial setting for around 25–40 h, the shield resistances keep relatively constant and are mostly below 30 MPa. In Figure 6(b), the continuous spatial-temporal evolution of shield resistance during 70 h in Panel LW63 of Cumberland mine are shown. The panel stops advance for 35 h as indicated in Figure 6(b). In this idle period, shield resistances increase in varying degrees. The temporal evolution of shield resistance during a typical shield supporting cycle under strong roof and medium-weak roof are plotted in Figure 7. The data for strong roof are the results during a roof-weighting period, during which the shield resistance increases abruptly after initial setting. The initial loading rate (first ∼5 h) is 0.083 MPa/min, after which the shield reaches the shield capacity, and thus the yield valve starts to open frequently. The initial loading rate (first ∼20 h) is 0.015 MPa/min for the case of medium-weak roof, the shield resistance tends to stabilize below the shield capacity after around 20 h.

In general, for strong roof, the shield resistance after initial setting during idle time depends on roof activity. If the main roof already breaks and starts to rotate, the shield resistance will increase rapidly. The shield resistance keeps at a high level during the whole idle time. The load increment and roof convergence both increase as the idle time increases. If the main roof does not break, the weight of immediate roof determines the shield resistance. In this case, the shield resistance keeps at a low level during the idle time. For the case of medium-weak roof, there is no obvious periodic weighting during the idle time. The shield resistance is dependent on the weight of overlaying strata, and its slow increase is mainly caused by strata splitting.

To consider the effect of retreating rate, the frequency distributions of TWAP in supporting cycle with duration shorter and longer than 2.5 h for Panel LW22304 with hard roof are shown in Figure 8(a). When the duration of supporting cycle is shorter than 2.5 h, TWAP shows a typical normal distribution with an obvious plateau at the right corner. When the duration of supporting cycle is longer than 2.5 h, the TWAP shows two distinct peaks, representing the load distributions in nonweighting and weighting periods, respectively. The left peak represents the maximum TWAP during the nonweighting period, which is nearly at the same value in comparison with that in the supporting cycle with duration shorter than 2.5 h, indicating that the load does not increase too much during the nonweighting period in supporting cycles with duration longer than 2.5 h. However, the right peak shows a TWAP as high as 40 MPa in the roof-weighting period. When the duration of supporting cycle is longer than 2.5 h, the supporting cycle with TWAP larger than 40 MPa (shield capacity is 47.2 MPa) occupies 23%. For the supporting cycle shorter than 2.5 h, TWAP of only 7.5% of the supporting cycle are larger than 40 MPa (Figure 8(b)). The load distribution of shield in Panel LW 63 with the medium-weak roof shows different characteristics. With the increase of the duration of the supporting cycle, the peak of TWAP distribution moves rightward (Figure 8(b)), indicating that TWAP decreases a little bit with increasing retreating rate. According to our study, the longwall retreating rate has a profound influence on roof loading. Although faster retreating rate cannot prevent the convergence of overlaying strata, it can maintain a good roof condition, achieving a good roof control quality.

4.2. Effect of Setting Pressure

The traditional view on the setting pressure of shield is that a higher setting pressure not only can help to reduce splitting of roof strata and thus improve the stability of roof, but also can prevent spalling of coal wall and reduce the roof damage in between the shield canopy and the coal wall. It is recommended that 0.6–0.85 times of shield capacity can be set as the setting pressure [27, 28]. However, the traditional view was proposed based on the low shield capacity several decades ago. Nowadays, with the increase of shield capacity, our data show that the average setting pressure is 20.18–25.82 MPa in all the 10 panels (Table 3), which is only approximately half of the shield capacity. Therefore, there are still several questions to be addressed regarding the setting pressure. Our previous work shows that the setting pressure has little influence on the load-increasing characteristics of shield after initial setting when there are no or few yielding events [24]. As shown in Figure 9(a), neighboring shields with different setting pressures follow a similar temporal evolution trend of shield resistance during a typical supporting cycle. Besides, TWAP is positively correlated with the setting pressure and a higher TWAP can be achieved by a higher setting pressure (Figure 9(b)). The setting pressure imposes minor effect on the pressure increment during the whole supporting cycle (Figure 9(b)). When the shield capacity is inadequate, yielding is frequent, and a higher setting pressure tends to increase the number of yielding (Figure 10(a)). Moreover, a higher setting pressure shortens the time required to reach the first yielding and increases the cumulative leg closure (Figure 10(b)).

4.3. Effect of Yielding

For a shield showing frequent yielding, the yielding characteristics are far more important than the shield capacity, whereas the mine managers and mining engineers usually take the shield capacity as the only parameter to assess a shield. They may hold the view that the frequent yielding of shields is a sign of inadequately designed capacity. In fact, even though the shield capacity reaches 2200 t in Shendong coal field, yielding still occurs frequently.

If the shield capacity is inadequate, the supporting load provided by the shield is not enough to prevent the overlaying strata from moving down, leading to the open of yield valve. According to the field data, the allowable leg closure before yielding is very limited because of the large stiffness of the shield. Normally, the leg closure before yielding is smaller than 10 mm except for cases with extremely low setting pressure (Figure 11). Moreover, the stiffness of shield drastically reduces during continuous yielding compared to that before yielding. As shown in Figure 11, the stiffness becomes 0.36 times of that before yielding. The reduced stiffness of shield leads to dozens of times of leg closure than that before yielding or even larger.

Hence, if yielding cannot be avoided, the yielding characteristics of shield determine the supporting characteristics of shield compared to the shield capacity. An ideal working condition of yield valve is that it can respond quickly to the yield load. That is, the yield valve can open immediately once yield pressure is achieved to expel a small quantity of hydraulic fluid and then close quickly. A desirable working condition of yield valve can ensure a small leg closure but present shield crushing from happening.

4.4. Effects of Shearer’s Cutting and Advance of Shields

Before and after shearer’s cutting and neighboring shields’ advance, the stable shield-roof interaction state is interrupted, leading to the rapid increase of shield resistance until a new stable state is reached. Field data show that the influencing period of shearer’s cutting and neighboring shield’s advance is in between about 20 mins after initial setting and about 10 mins before shield advance. During this period, the load increasing rate (0.2∼3 MPa/min) is normally dozens of times of that during relatively stable period (less than 0.1 MPa/min).

Previous studies conducted by the authors show that under the independent influence of shearer’s cutting (Figure 12), i.e., without neighboring shield’s advance, the load increase in the short period preceding shield advance is related to the position of the shearer’s (the distance between the shield and the shearer’s front drum) by the following function [24, 29]:where is the final value after pressure increase and and are constants.

Under the combined influence of shearer’s cutting and neighboring shield’s advance (Figure 12), i.e., the neighboring shields are advanced after shearer’s cutting, the load increment in the short period preceding shield advance is related to the position of the shearer’s (the distance between the shield and the shearer’s front drum) by a quadratic function [29]:where , , and are constants.

In addition, the geological condition and the shield capacity can impose additional influences on the load increasing characteristics during shearer’ cutting and neighboring shield’s advance. For panels with a strong roof, the roof activity is critical for the load-increasing characteristics during shearer’ cutting and neighboring shield’s advance. As shown in Figure 13(a), the cantilever beam is short, and the pressure increment after initial setting and before advance is not obvious (stage a in Figure 13(a)). With the increase of the cantilever beam length, the equilibrium state of roof is interrupted and the loading rate before advance increases (stage b in Figure 13(a)). Afterwards, the load increasing rate during relatively stable period increases dramatically until the frequent open of yield valve (stage c in Figure 13(a)).

When it comes to weak roof equipped with shields of sufficiently high capacity (Figure 13(b)), the pressure increment after initial setting and before advance is large, which can be 5–25% of the shield capacity in the case of Figure 13(b). Moreover, the load increasing rates after initial setting and before advance are similar. This is attributed to the fact that the weight of overlaying weak roof is the load on the shield. When the shield capacity is adequate, the shield resistance increases rapidly during shearer’s cutting and neighboring shield’s advance (Figure 13(b)). For inadequate shield capacity, yield valve opens more frequently (Figures 13(a) and 13(c)). However, it is hard to distinguish this difference from the shield resistance and loading rate. The difference is clear when it comes to leg closure (Figure 13(c)), the rate of which (normally 1 mm/min) can be one magnitude higher than that before shearer’s cutting.

5. Preliminary Application of SSRI in Ground Control

Building on the insights gained on shield-roof interaction from field data interpretation, we present the preliminary application of the SSRI system in ground control in two aspects, i.e., early warning of severe periodic roof weightings and assessment of the supporting quality of shield.

5.1. Early Warning of Severe Roof Weightings

The periodic roof weighting in longwall panel is temporally nonuniform along the dip direction. The real-time monitoring of leg pressure of all shields in the longwall panel provides the dataset for SSRI to extract the characteristic parameters of shield associated with roof weightings to construct the warning model of roof weightings. By detecting the few shields that are experiencing roof weighting, we can predict the arrival of an incoming large-scale roof weighting (Figure 14).

Specifically, we first chose the characteristic parameters to evaluate the loading degree of each shield-supporting cycle, including the load-increasing rates during the initial setting period, relatively stable period, and before shield advance, the ratio of TWAP and shield capacity, the yielding characteristics, leg closure rate, and time duration of each supporting cycle. The decision tree algorithm has been employed to evaluate the loading degree in the current supporting cycle by identifying the characteristic parameters in the last supporting cycle.

Upon the completion of the warning model of roof weightings, we establish a roof-weighting index to classify the intensity of roof weighting into 5 levels (Figures 15 and 16). Once the index exceeds the critical value, an early warning signal will be sent out.

5.2. Assessment of the Supporting Quality of Shield
5.2.1. Assessment of Yield Valve Open

As discussed above, the working condition of yield vale is critical when the shield capacity is inadequate. However, field observation shows that the yield valves are not maintained properly, resulting in a low stiffness of shield. Taking the Yongcheng mine as an example, we discuss the existing issues on yield valve.

(1) Higher Open Pressure and Lower Close Pressure. If the yield valve opens before yield pressure, we do not take full advantage of the shield capacity. In other cases, the yield valve may keep closed beyond yield pressure, which is dangerous because high fluid pressure inside the cylinder can probably cause shield damage. As shown in Figure 17(a), more than half of the shields in Panel 203 in Yongcheng mine are malfunctioning.

(2) Delayed Response of Yield Valve. The yield valve may also fail to close rapidly after it opens. In this case, the long duration of yield valve open decreases the stiffness of the shield, and thus the roof moves down significantly. When there are severe roof weightings or longer supporting cycles, the shield may experience crushing. Most of the yield valves keep open for more than 1 min in most of the yielding cycles (Figure 17(b)), leading to a leg closure more than 1 mm in 65.2% yielding cycles (Figure 17(c)), and the pressure reduction are more than 2 MPa (Figure 17(d)).

5.2.2. Diagnosing Fluid Leakage

Except for the existence of rock debris, soft base, and/or roof, the sealing leakage of leg cylinder is the main cause of the decrease of shield resistance.

The decrease of shield resistance caused by two different degrees of fluid leakage is shown in Figure 18, with the first case experiencing a minor fluid leakage (Figure 18(a)) while the second experiencing a rapid fluid leakage (Figure 18(b)). In the case of minor leakage (Figure 18(a)), the decreasing rate of shield resistance is relatively constant and small at 0.01–0.1 MPa/min. The leg has a certain residual resistance before advancement. The decrease of shield resistance caused by minor fluid leakage is extremely small and thus difficult to detect. For the case of rapid fluid leakage (Figure 18(b)), the decrease of shield resistance can be 0.5–1 MPa/min or even larger. The shield losses its resistance shortly after initial setting. If the automatic pressure compensation function has been set, the continued leakage urges the system to supplement hydraulic fluid into the cylinder, which increases the burden of hydraulic system.

For a single leg, a minor leakage can evolve to a rapid leakage due to repeated loading and unloading. For the two legs of a shield, as shown in Figure 19, one leg may be experiencing minor leakage first, and another leg carries most of the loads on the canopy. In this way, the yield valve of another leg frequently opens and closes, and finally sealing failure occurs in this leg. The shield ultimately losses its supporting capacity.

When a shield is experiencing rapid fluid leakage, the roof loading on it transfers to its neighboring shields. The roof is under a “high-low-high” vertical supporting load, which induces shear stress in the roof strata in between these neighboring shields. If the roof is experiencing a severe roof weighting, the roof strata deteriorates and even roof fall may occur. The SSRI system can be employed to detect the fluid leakage of shield. During the 74 h of retreating in Panel 22306 in Bulianta mine, 16 out of 123 shields experience fluid-leakage problem (Figure 20).

6. Conclusions and Prospects

In this study, we developed a powerful software package SSRI, to analyze tons of monitoring data on leg pressure and leg closure of shields in 1.6 million supporting cycles in 11 longwall panels of coal mines in China and the U.S.

The results show that the effects of time (idle time and retreating time) on supporting characteristics of shield are dependent on roof conditions. Setting pressure imposes little effect on the supporting characteristics of shield when there is adequate shield capacity, while for inadequate shield capacity, the first yielding is advanced, and the number of yielding and the leg closure increase by higher setting pressure. The supporting capacity of shield depends on its yielding characteristics when it experiences frequent yielding. The delayed response of yield valve lowers the stiffness of the shield and thus its supporting capacity. Overall, the designed shield capacity is not the only factor that should be considered when choosing the shield model; the time-related factors, the geological condition, the setting pressure, yielding characteristics, and mining method are to be considered collectively.

Due to the lack of knowledge and techniques to determine the exact occurrence characteristics and behavior of the rock strata in their natural state, the coal extraction proceeds under not well-known geological conditions. The monitoring and analysis of the leg pressure and leg closure of shield provide an avenue to continuously understand the ever-changing shield-roof interaction with the proceeding of mining. This study only presents our preliminary results, and there is a lot of work to be done. For example, determination of time-dependent roof movement at longwall faces and its implication in shield capacity determination is to replace the empirical equations for shield-capacity determination. In addition, the evolution of each individual shield and the roof-loading transfer and redistribution characteristics in neighboring shields can be used to predict roof fall and coal spalling. As another example, for the intelligent mining strongly advocated by the Chinese government and the business community, we may achieve self-adaptive control of shield by deep understanding of the complicated shield-roof interaction.

Data Availability

The data used to support the findings of this study are included within the article.

Conflicts of Interest

The authors declare that there are no conflicts of interest regarding the publication of this paper.


The authors would like to express their sincere appreciation to Prof. Syd Peng (West Virginia University) and Dr. Perter Zhang (Alpha Natural Resources) for their contribution in this study. This study was supported by the National Natural Science Foundation of China (grant number 51604267), the State Key Laboratory of Coal Resources and Safe Mining CUMT (grant number SKLCRSM15X05), and the Open Projects of Research Center of Coal Resources Safe Mining and Clean Utilization, Liaoning (grant number LNTU16KF01).