1. Introduction
Coordinated by KOMAG Institute of Mining Technology (ITG KOMAG) and financed by RFCS (Research Fund for Coal and Steel) the INESI project (Increase Efficiency and Safety Improvement In Underground Mining Transportation Routes) aims to improve work efficiency and crew safety in underground hard coal mines. One of the main objectives of the project is to increase speed while improving safety in underground mining transport. Increasing the speed of suspended trains is related to the necessity of developing an effective braking system [
1,
2] and monitoring transport routes.
Monitoring of engineering structures is a very important issue, both in terms of proper functioning of the monitored object, as well as safety of use and impact on the environment. Safety issues are especially important in the case of mining infrastructure. Science is constantly trying to improve technical solutions, proposing the introduction of new technologies to monitoring, such as IoT (Internet of Thing) [
3,
4], artificial intelligence (e.g., machine learning) [
5,
6].
In publication [
7], the authors present a system for monitoring lateral wall deflection in a mine based on strain measurement, by means of designed components using film strain gauge bridges. Based on tests, it is shown that the accuracy and resolution of the proposed method is sufficient for this type of measurement. The designed system is widely applicable to provide up-to-date data for safety assessment to prevent unexpected destruction of underground excavations as well as other engineering structures.
There are already known ways of monitoring the geometry of powered roof support geometry. The publications [
8,
9] presents the results of testing the geometry system of mechanized support sections using inclinometers, which were made in accordance with the requirements of the ATEX directive. One of the challenges related to controlling and monitoring parameters of powered roof support section is mapping its geometry and mutual arrangement of individual elements. Tests of the system were carried out on a real object in laboratory conditions. Test results will be used during implementation of the system in real conditions.
In the case of arch supports in mine tunnels with suspended monorails, the need for continuous monitoring has not been noted to date. Measurements of tunnel dimensions are made by qualified personnel only during tunnel construction and are intended to ensure proper routing of the excavation. Increasing the speed of suspended monorail in mines forces monitoring of the geometry and loading of the supports, as deformation of the supports can cause deformation of the monorail track and create a risk of accident. In addition, the increased speed of the train itself can cause a change in geometry, such as when clamps are already at the limit of the loads that can cause the support to slide and lower.
The project proposed a measuring system, dedicated to roadway support, the idea of which is presented in
Figure 1 [
10].
The monitoring system consists of multiple identical measurement modules connected in series. Each measuring module is equipped with draw-wire sensors and vibrating wire strain gauges. The information is successively transferred from one module to another, and at the end of the line it is collected by the recording module. This topology minimizes the amount of wiring needed and, in the case of a wireless solution, allows the monitoring system to extend its range. The collected data is read by an external computer system, from one module. The hardware module is supported by software designed to record, analyze measurement data and generate warnings about the risks associated with the deformation of the route of the train, based on the developed expert module. The change of the geometry of the roadway supports, caused by the slides on the clamps and by the abnormal load, can cause the lowering of the suspension and as a result the change of the angle between the rail sections, on which the monorail moves. Permissible values of sling lowering are included in technical conditions for suspended monorails.
KOMAG developed the concept of the measuring system, which was presented in detail in an earlier publication [
10]. During the development of the concept, modifications were introduced, consisting in changing the location of the vibrating wire strain gauges (
Figure 2).
The concept is to use two measuring modules (on both sides of the arc case). Directly at each module, vibrating wire strain gauge is attached (by gluing) to the profile. The sensor is additionally protected by a shield and, if necessary, can be protected by filling it with resin. The draw-wire sensor will be built in the same shield, and the measuring wire will be attached to another part, allowing the displacement measurement between the parts of the roadway support.
The aim of the research was to verify the proposed measuring method a vibrating wire strain gauges and to verify the work of the measuring module. The verification of the measurement method is based on the comparison of the results obtained from the measurements by two methods—vibrating wire strain gauges and film strain gauges. Film strain gauges are commonly used to measure strain in monitoring different engineering structures [
11,
12]. In industry, they are often used as main measurement systems, while in scientific publications it is common to find the use of strain gauges as a reference source [
13,
14].
The following sections present the proposed measurement method, describe the construction of the hardware model of the measurement module, discuss the principles of determining the load from the wire vibration frequency, and present and analyze the test results.
3. Results
As part of the INESI project, the developed measurement method was tested. Test measurements of deformations of frame of arch support, type ŁPP10/V29/4/A/I, with the use of vibrating wire strain gauges, were carried out on the test stand of the Central Mining Institute. The testing methodology is based on the PN-G-15000-05:1992 Standard, which defines the method of loading the frame of support, presented in
Figure 10. Active and passive forces are applied to the support by hydraulic cylinders, powered from a pumping unit. Asymmetric loads on the support frame can also be applied during the tests, e.g., by independent control of the cylinders [
23].
The arc support consists of several steel arches, connected with friction locks. Until the moment when static friction in the lock is able to counteract the movement of arches (under the pressure acting on the support), the construction works as a rigid support, showing sensitivity to loads only within the limits of elastic deformation of the material. When the forces applied to the support increase and are able to overcome the friction forces in the locks, they are suddenly slipped (
Figure 11). The rock mass deformation process is slow, which leads to a load decrease on the support and, at some point, the friction forces in the lock are again able to counteract the displacement. Until the next state of balance is exceeded, the support will operate as rigid [
24].
Tests were performed on a full-size arc support installed in a test stand. The arch support frame is loaded from the roof side by three hydraulic cylinders. The response from the side wall was mapped using 6 hydraulic cylinders according to the diagram shown in
Figure 10. The load applied to the structure is estimated by measuring the pressure in the hydraulic system.
During the measurement, slides were recorded in the joints between the roof arch and the side wall arch, using draw-wire displacement sensors. An additional displacement sensor was provided to measure the extension of the F5 cylinder. The deformation of the profile was recorded using two vibrating wire straining gauges mounted on both sides of the arc support. Sensor data was collected by using a measuring module developed within the project (
Figure 12). The measurement module records the frequency of vibration of the measuring wire in the sensor. This value, must be converted to strain, using the relationships, provided by the sensor manufacturer [
15].
In order to verify the measurement method, film strain gauges (
Figure 13) were additionally applied to the arch of arc support. It is the most common solution for measuring tensile, compressive and stress forces. Resistance film strain gauges are characterized by high sensitivity and accuracy and resistance to vibration and shock. The biggest disadvantage of this solution is their fragile construction (on foil) and the associated difficult and lengthy installation. The use of solutions in the form of force sensors, in turn, is associated with a significant expense, especially in the case of ATEX solutions. Data from strain gauges were recorded with the SPIDER 8 measurements amplifier (0.2% ampli-fier class) and CATMAN EASY software.
Correctly designed support for roadways behind the front of longwall will maintain their functionality for the entire intended period of use. During the design of the support for roadways behind the front of longwall, an important issue is to accurately predict the convergence parameters that will occur during the expected service life [
24]. Such on approach will permit the choosing of the right size of the roadway support frame. The deformation of roadways makes the mines to carry out reconstruction work. This is the only way to restore the required profile dimensions of a roadway. Reconstruction of roadways is also a time-consuming and costly operation, additionally disrupts the operation of the roadway. In the INESI project, additional factors, influencing the stability of roadway supports, were analyzed, such as the increase of speed of suspended trains.
The place of installation of vibrating wire stain gauge is very important. Preliminary research, conducted using a 30 cm section of the V29 profile, was loaded on a test stand in KOMAG. The vibrating wire strain gauge was mounted by welding the handles on the side of the profile (
Figure 14). This location of the sensor was dictated by concern for its mechanical protection.
During preliminary testing, inconclusive results were recorded with the vibrating wire strain gauge. This could be due to the fact that the profile was not cut perfectly and therefore, there was more load on one of its edges, causing uneven deformation. After preliminary research, the most suitable position for an installation of sensors on the arch surface was selected. Sensors were mounted on the thickest, flat wall of the profile (
Figure 15). Such a location of the sensor minimizes the influence of improper conditions, i.e., support the profile on edge, side bending forces, etc.
In addition to the vibrating wire strain gauges, both sides of the arch support frame were additionally mounted with film strain gauges for a later comparison of results.
The data from the wire strain gauges, recorded during the tests, are compared and presented in
Figure 16 and
Figure 17, which show that the recorded deformation and slide values are very similar on the joints of the support.
Based on the results of the research mentioned above, a comparative analysis of both measurement methods (i.e., the method of arch support deformation measurement using film strain gauges and vibrating wire strain gauges) was carried out using Spearman correlation and the Bland and Altman method [
25].
Figure 18 and
Figure 19 show dispersion diagrams with a regression line for selected pairs of variables, i.e., arch support deformation (left and right side of arch) measured by film strain gauges and vibrating wire strain gauges, respectively.
The graphs show a strong positive correlation between the two measurement methods, which is confirmed by the values of the Spearman coefficient. For the data presented, the values of this coefficient are, respectively: R = 0.951 (
p < 0.05) (
Figure 18) and R = 0.986 (
p < 0.05) (
Figure 19). The null hypotheses are that the measurements made by the two methods (independently for the left and right side of the arch) are not linearly related. Very small probability values (
p-values are less than the significance level α = 0.05) allow us to conclude that measurements with film strain gauges and vibrating wire strain gauges are correlated. However, a high correlation does not mean that the two methods are compatible because R measures the strength of the relationship between two variables, not the compatibility between them. In view of the above, in the next step of the analyses, for the purpose of assessing the differences between the measurement methods, the data obtained were presented in a graph of the difference between the methods in relation to their mean (the so-called Bland-Altman graph). This plot (also known as the Tukey mean difference plot) is most commonly used in chemometrics and biostatistics. It represents the degree of agreement between two different samples or indicators and therefore one of its main applications is the comparison of two clinical indicators, each of which is subject to some measurement error [
26,
27,
28,
29]. It can also be used to compare and assess the equivalence of quantitative measurement techniques. The plot of difference against the mean also allows the examination of the possible relationship between the measurement error and the true (real) value. The true value is not known and the average of two measurements is the best estimate that can be obtained. It would be a mistake to plot the difference against either value separately because the difference will be related to each, a well-known statistical artefact.
An assumption of the Bland-Altman limits of agreement is that the differences are normally distributed. Statistical tests have been used to determine if the distribution is normal, since in some cases normality cannot be determined simply by observing the histogram plot (
Figure 20 and
Figure 21). A test for normal distribution (Kolmogorov-Smirnov test, Shapiro-Wilk test, D’Agostino-Pearson) have been carried out, for the hypothesis that the distribution of the observations in the sample is normal (if
p < 0.05 then rejected normality) (
Table 1 and
Table 2).
Figure 22 shows a comparison of the deformation of the left arch of the support measured by a film strain gauge and with a vibrating wire strain gauge. The mean difference is 0.0064 with 95% confidence interval 0.001 to 0.012. Thus, wire strain gauge tends to give a higher reading, by between 0.001 and 0.012. Despite this, the limits of agreement (−0.0605 and 0.0732) are small enough for us to be confident that the new method can be used in place of the foil strain gauge method. Limits of agreement (LoA) estimate the interval within which a proportion of the differences between measurements lie.
Table 3 shows all the Bland and Altman plot statistics.
Figure 21 shows a comparison of the deformation of the right arch of the support measured by a film strain gauge and with a vibrating wire strain gauge. The mean difference is 0.0265 with 95% confidence interval 0.024 to 0.029. Thus, wire strain gauge tends to give a higher reading, by between 0.024 and 0.029. Despite this, the limits of agreement (−0.0073 and 0.0602) are small enough for us to be confident that the new method can be used in place of the foil strain gauge method.
Table 4 shows all the Bland and Altman plot statistics.
In a Bland and Altman plot system the differences can be also expressed as percentages of the values on the axis. This option is useful when there is an increase in variability of the differences as the magnitude of the measurement increases. Figures 24 and 25 represent the same data as
Figure 22 and
Figure 23, plotted as percentage of differences. The bias (mean difference) is 5.44% and 17.05%, and it is higher for the very low values. The agreement limits are from −68.35% to 79.22% (
Figure 24) and from −41.78% to 75.88% (
Figure 25).
On this basis, it was concluded that the vibrating wire sensors can be used to assess the change of the roadway support load (
Table 5).
An estimation of the support geometry requires a measurement of the friction lock slide.
Figure 26 shows the time waveforms of the slide values, recorded by draw-wire sensors, which are placed on both sides of the support in front of the locks in such a way that the wire runs from the sensor to the anchorage point behind the lock. The third sensor was mounted on a stand and measured the extension of the main load actuator.
The displacements between arch support parts due to slides during testing can be seen in
Figure 27.
Displacements can in reality be much larger, which will lower arc support by up to several tens of centimeters.
4. Discussion
The research and analysis of the test results achieved during the project allows to conclude that the vibrating wire strain gauges, which are increasingly used to monitor structures, can be used to measure the deformation of the arch support. The results obtained from vibrating wire strain gauges and film strain gauges are similar. In the case of tests carried out in-situ, it is important to take into account the difficulties that may occur on a larger scale.
The measurement of the sliding distance between the elements of the support with the use of vibrating wire strain gauges, applied in the target measurement system, allowing the estimation of the changes in the support geometry. Collected data enable the analysis of possibility of mounting the vibrating wire strain gauges on the support installed in the roadway, which already shows some signs of deformation. Moreover, the results, achieved during the tests, are a valuable contribution to the expert system, which will be used to assess the operating conditions of the support in terms of increasing the speed of suspended trains.
Due to the way the roadway support is operated and the elastic and plastic deformations occur in it, a determination of the exact geometry of the support in real conditions is an impossible task. However, the aim of INESI project is not so much to measure the geometry accurately but to alert that the slide has occurred or (basing on the load measurement) may occur soon. This makes it possible to send information to the train operator that there may be obstacles on the route.
Directions for further work should focus on teaching the expert system using results from tests of different types of arch support, under different loading conditions (static and dynamic). Currently, the final results during testing were converted from measured values of wire vibration. It is planned to implement into the software an automatic conversion of vibration frequency, to deformations and forces acting in the arch support structure. This will make it possible to directly analyze current results and compare them with reference data and generate appropriate warnings or alarm conditions.
In the process of work, important knowledge was acquired in the field of application and maintenance of vibrating wire strain gauges in mining conditions. The functioning of the designed and constructed measuring module, complying with the requirements of the ATEX directive, was verified. Comparable test results recorded by sensors with a vibrating wire strain gauges and film strain gauges confirmed the correctness of the monitoring system. The analysis of the test results of the sensors placed in different places of the steel profile allowed selecting the place where the vibrating wire strain gauges is the least sensitive to the disturbance of the measurement signal. As part of the testing, the vibrating wire strain gauge mounts were welded to the steel profiles of the arch support. However, this type of mounting would not be acceptable in underground mines, so it was determined that gluing the mounts was an equally good mounting method (provided the adhesive recommended by the sensor manufacturer was used).