Zhenjun Shn*, Ting Ren Jn Nemcik Gunzheng Wu Ling Zho

a School of Civil, Mining, Environmental and Architectural Engineering, University of Wollongong, Wollongong, NSW 2522, Australia

b Azure Mining Technology, CCTEG, Chatswood, NSW 2067, Australia

Keywords:Standing support Underground spaces FRP Tailings and coal wash rejects Design-oriented model

A B S T R A C T Twenty-seven specimens were tested to investigate the uniaxial compression behaviour of an innovative standing support for underground space applications.The innovative standing support consisted of an external fibre reinforced polymer (FRP)jacket and the infill column made of cementitious grout, tailings and coal wash rejects.Effects of the FRP layers number and water to the cementitious grout (w/g) ratio were evaluated.Test results indicated that lower w/g ratios produced stronger infill columns.With FRP confinement, the standing support demonstrated strain-hardening loading characteristics with a significant improvement in both strength and ductility.The highest strength and strain of the specimens achieved was 58.4 MPa and 11.8% respectively.Compared with the unconfined specimens, the confinement with four FRP layers increased the specimen strength and associated strain up to 3.6 and 27.0 times respectively.A correlation between the compressive strength of the infill material and the ultrasonic pulse velocity was also investigated.Furthermore,a simple design-oriented model was proposed to predict the peak strength and the corresponding strain of the innovative standing support.

1.Introduction

Roof support has always been one of the most challenging practices in underground coal mining operations.Unstable roof strata can cause roof falls and other associated problems, resulting in a disruption of mine production and safety issues.Typically, underground coal mine roof support systems include primary support such as rock bolts and secondary support such as cable bolts,welded steel mesh,shotcrete and standing supports.Standing supports have been widely used in longwall tailgates in underground coal mines.According to Barczak et al.[1], standing supports can be classified into 4 types in terms of their loading features as shown in Fig.1: (a) non-yielding, (b) constant yielding, (c) load shedding or strain softening,and(d)load increasing or strain hardening.The representative supports of the 4 types are concrete crib,Can support, pumpable roof support and wood crib respectively.Concrete crib has a large load carrying capacity,but it has a limited deformation capacity.Due to its non-yielding characteristic, the concrete crib loses its load bearing capacity completely after failure.It may also experience premature failure when a high roof deformation exists.The Can support is structurally a steel tube with infill material inside.While the Can support has its deformation capacity as high as 50%, it needs to be topped to create roof contact, compromising the stiffness of the support system [2].Another disadvantage of the Can support is its potential to cause gas explosion by sparks when the shearer drum cuts the supports[3].Pumpable support consists of a fabric bag which is normally hung from the roof and the cementitious grout is pumped into the bag.Pumpable support is generally much stiffer than the Can support, but it experiences significant load shedding after peak load [4].The load shedding event is attributed to the brittle infill material[5].Compared to the other three standing supports,wood crib is a soft support[2]and its installation is time consuming[4].

Fig.1.Loading characteristics of typical standing supports [1].

Although conventional pumpable standing supports have been successfully used in underground mines,research for novel standing supports with superior loading characteristics is on-going.Due to its large strength-to-weight ratio and high corrosion resistance[6], fibre reinforced polymer (FRP) has found wide applications in civil engineering such as retrofitting existing reinforced concrete columns [7,8] and constructing new FRP confined concrete columns [7].With the confinement provided by the FRP, both the strength and deformability of the concrete can be improved[9,10].From the design aspect of standing support, maintaining the load carrying capacity over a large deformation is a desirable characteristic.Because part of the roof-to-floor convergence in underground mine roadways is unstoppable or uncontrollable,the standing support without sufficient deformability may experience premature failure[11].In light of the successful application of FRP in civil industry,an attempt was made at the University of Wollongong to make a FRP standing support for use in underground spaces such as underground mines [3].That standing support was a FRP tube that was infilled with high-water cementitious grout and coarse aggregate material.Upon compressive loading,the infill material expands and induces tension in the FRP.The tensile resistance, in turn, produces confining pressure to the infill material, resulting in a significant improvement in both strength and ductility.Further studies have also been conducted to investigate the effects of various factors on the performance of the FRP hybrid standing support, including: types of FRP [12], infill material composition such as screened and unscreened coal rejects[13], FRP thickness, confining jacket configuration and strength of infill material [14].

Due to the growing population, booming economy and rapid urbanization, the solid wastes have been increasing significantly,which becomes a challenge to the world as it brings adverse impacts on human health, natural resources and the ecosystem[15].Many studies have been conducted to evaluate the feasibility of recycling these waste materials, aiming to reduce greenhouse emissions which has been attracting increasing attention[16].Partial replacement of cement in normal concrete with solid wastes such as waste glass powder, coal rejects powder and fly ash was investigated in a study [17], it was observed that the sustainable concrete with the inclusion of the wastes had greater compressive strength than the normal concrete when exposed to sodium sulfate solution for 22 months.A new regenerated binding material(RBM)made of steel slag power,ground blast furnace slag,fly ash,carbide slag,gypsum,red mud and coal rejects was found to be able to perform slightly better than the ordinary Portland cement in terms of 28-day compressive and flexural strength [18].Experiments were carried out to assess the effect of fibers on the behaviour of recycled aggregate concrete(RAC),test results indicated that inclusion of the fibers in the RAC did not increase its compressive strength,but the durability of the RAC was improved [19].

This study seeks to investigate the potential of recycling two types of mine waste, namely coal wash rejects and tailings.Coal wash reject is a typical mining industry by-product.Emplacement of the coal rejects has become a challenging issue in Australia due to its large volume and environmentally unfriendly nature [20].Tailings are another type of mining waste, which have an adverse influence on the environment and even caused fatality incidents[21].One of the feasible methods to address mining waste problem is to turn them into products for structural support in mines [22].Studies have been conducted to evaluate the performance of tailings [23] and coal rejects [13] as backfill materials in coal mines.However, investigations on combining both coal rejects and tailings into a single product,such as FRP standing support,have been limited.The current study is an attempt in this direction.

Current research on the innovative FRP standing support at the University of Wollongong investigated the feasibility of incorporating both coal wash rejects and tailings into the infill material.Potential applications of this FRP standing support aim at rock support in underground spaces, including but not limited to underground coal mines.Successful application of this standing support not only helps to lower the cost but also benefits the environment.A series of laboratory tests were carried out to evaluate the uniaxial compression performance of the innovative standing support with various material water ratios of infill column and number of FRP layers.In addition,a design-oriented mathematical relationship was also proposed to predict the peak strength and the corresponding strain of the novel standing support.

2.Experimental program

2.1.Proposed specimens

This study attempts to assess the behaviour of a novel FRP column standing support for use in underground spaces.The innovative standing support includes two major parts: (1) the infill column made of cementitious grout,coal wash rejects and tailings and (2) external FRP jacket which provides confinement to the infill column upon loading.The infill column was 150 mm in diameter and 300 mm in height.As listed in Table 1,27 specimens in 9 groups were prepared.Each group consisted of 3 nominally identical samples.The grout, coal rejects and tailings were mixed at a constant ratio for all the infill columns.Three different water-togrout(w/g)ratios of 0.8,1 and 1.2 were used to cast infill columns with various strengths.To investigate the influence of FRP thickness on the performance of the standing support, three types of FRP configurations were applied to the columns: (1) without any FRP confinement, (2) with two layers of FRP confinement and (3)with four layers of FRP confinement.For easy reference,each specimen group was given an abbreviated name.The group name started with the letter ‘R’ followed with the number indicating the w/g ratio, then by a letter ‘F’ followed with the number of FRP layers.For instance, R0.8F2 represents the infill cylinder had the w/g ratio of 0.8 and it was confined by 2 layers of FRP liner.

2.2.Materials

The infill cylinder was made of cementitious grout as the binder, tailings as the fine aggregate and crushed coal wash rejectsas the coarse aggregate with a certain mix ratio.The cementitious grout was provided by a local Australian company while tailings and coal wash rejects were provided by local Illawarra mines.The tailings and the coal rejects were kept in a 105 °C oven for 24 h to ensure consistent moisture content.The material was then crushed into pre-determined sizes.Fig.2 illustrates the materials used to cast the infill column.The coal reject had a nominal maximum size of 10 mm,its particle size distribution is shown in Fig.3.The glass fibre fabric (Fig.4e) and epoxy resin were provided by a local supplier.The fabric has 90%fibre by mass in the major direction and 10% fibre at perpendicular direction.As can be seen from the bottom right part of Fig.4e,some fibre strands in the direction perpendicular to the major direction are intentionally pulled out to demonstrate the woven configuration of the fibre sheet.Even though the FRP is not recyclable, it is believed that its influence on the underground environment is limited.

Table 1 Description of the groups of the specimens.

Fig.2.Materials used to cast the infill column.

Fig.3.Particle size distribution of the coal wash rejects and the Australia Standard recommended grading for 10 mm coarse aggregate used in concrete [24].

Fig.4.Specimen preparation.

2.3.Specimen preparation

Fig.4 describes the process of specimen preparation.The grout,tailings and coal rejects with a certain volume ratio were mixed evenly (Fig.4a), pre-determined amount of water was added and mixed until the mixture was of uniform colour and consistency(Fig.4b).Slump tests were conducted on selected mixtures to determine the workability of the infill material.The mixture was then cast into the steel cylindrical moulds.Vibrator was used to reduce air bubbles during casting.The top of the mould was covered with a thin plastic sheet to preserve the sample moisture(Fig.4c).The cylinders were cured in the steel mould for 24 h before removing and placing the samples in a humidity room for 2 weeks(Fig.4d).The nominated cylinders were then warped with FRP liners (Fig.4f).Note that for all the FRP confined cylinders, a 150 mm FRP overlapping zone was applied.All specimens were cured for 28 d and capped to ensure uniform loading before testing(Fig.4g).Fig.4h shows the specimens ready for testing.

2.4.Test procedure

Immediately prior to the compression test, ultrasonic pulse velocity (UPV) test (Fig.5a) was conducted on the infill columns to evaluate the correlation between the compressive strength and the UPV of the infill material.Three strain gauges were attached at the mid-height of the FRP surface of the standing supports to measure the circumferential strains of FRP during the compression test.When installing the strain gauges the overlapping zone was avoided.Two linear variable displacement transducers (LVDTs) were installed for each specimen to record the axial deformation.A 500-tonnes compression testing machine was used to load the samples.A 5 kN preloading was applied for all specimens to eliminate the slack between the surfaces of the columns and the loading plates.Displacement control mode was selected at the loading rate of 0.6 mm/min for all tests.The strains, loads and deformations were recorded simultaneously by a data logger.Fig.5b illustrates the compressive test set-up.

Fig.5.Setup of the ultrasonic pulse velocity test and the compressive test.

3.Results and discussions

3.1.Workability

Slump test was conducted to evaluate the workability of the infill material with various w/g ratios.As all infill mixtures were made of the same ingredients with the same mix ratio but three different w/g ratios, the mixtures can be classified into three groups in terms of w/g ratio.Each group includes nine specimens which can be further divided into three subgroups depending on the confinement applied.It is believed that the workability of the mixtures in each group should be the same or similar.Therefore,instead of evaluating the workability of every mixture in the group,only one mixture in each subgroup was subjected to the slump test.Test results are presented in Table 2.As expected,the workability of the infill material improved with increased w/g ratio.The slump was 35 mm when the w/g ratio was 0.8, it increased to 160 mm and 220 mm at the w/g ratios of 1.0 and 1.2 respectively.

3.2.Ultrasonic pulse velocity (UPV) test results

The ultrasonic pulse velocity test is a non-destructive testing method which has been successfully used to assess the quality of concrete materials [25].To evaluate the specimen quality of the infill columns, all infill cylinders were subject to the ultrasonic pulse velocity (UPV) test at the curing time of 28 d and prior to the uniaxial compressive test.Note that one of the plain infill columns failed prematurely during the compression test due to eccentric loading and therefore these results were omitted.The UPV test results are presented in Table 2.It is clear that the UPV was consistent in the infill material with the same water-to-grout(w/g)ratio,indicating that the consistency of infill material quality was achieved.The UPV of the infill material was inversely proportional to the w/g ratio, it dropped from approximately 2720 m/s to around 2250 m/s when the w/g increased from 0.8 to 1.2.This is because the volume of capillary voids in the concrete increased as a result of growing w/g ratio, which reduces the velocity of ultrasonic pulse passing through the concrete [26].Since the UPV is a function of the material density, it can also be correlated with the compressive strength of the concrete [27].Based on the test results shown in Fig.6, an exponential relationship between the UPV and the compressive strength of the infill column at the 28 d curing is presented here:

where fcois the compressive strength after curing for 28 d;and v the UPV, m/s.

3.3.Failure modes

Fig.7 shows the typical failure modes of the unconfined and FRP confined columns.The unconfined columns experienced brittle compressive failure and there was no post failure behaviour recorded.The FRP confined columns underwent stress hardening process when the expansion of the infill material in the lateral direction was large enough to activate the confinement of the FRP.After reaching the maximum stress, the FRP ruptured due to tension in the circumferential direction.Localized FRP rupture was observed either at the upper,the middle or the bottom section of the FRP tube.The circumferential rupture strain of the FRP was measured at around 0.8%, which was lower than its ultimate tensile strain of 1.4% [12].The actual hoop rupture strain of the FRP in confining concrete column has been reported smaller than its ultimate tensile strain in many studies[7,28,29].The main reasons are: (1) the curvature of the FRP jacket which leads to reduced strain capacity of the FRP; (2) the deformation localization of the cracked concrete and (3) existence of an overlapping zone, which result in non-uniform strain distribution in the FRP [28].It is also worthwhile to note that epoxy resin cracks were observed prior to failure in the FRP with the compression induced resin cracking producing noise.

3.4.Axial stress-axial strain behaviour of the specimens

Fig.8 illustrates the axial stress-axial strain curves of the specimens.The axial stress was the calculation based on the recorded compression load and the cross-sectional area of the column.The axial strain was obtained by dividing the average of the two LVDT readings by the original height of the specimen.The unconfined concrete cylinders had monotonic ascending shape of stressstrain curves before failure.In contrast, the stress-strain curves of FRP confined specimens were bilinear ascending in shape prior to peak strength, indicating that these infill columns experienced strain hardening as a result of the confinement provided by FRP.Unlike the plain infill cylinders which broke apart and were unable to resist any load after failure, FRP samples were still able to provide some variable levels of post peak residual strength.This was because the FRP confined cylinders were able to maintain their integrity to some extent after the localized rupture of the FRP.

Note that the slope of the stress-strain curve of plain infill cylinders was similar to that of the first ascending portion of FRP confined specimens, irrespective of the number of FRP layers.This was because FRP confinement was passive when not loaded.The FRP generated confining pressure only when the lateral expansion of the infill material occurred.In the first ascending portion, the dilation of the inner column was too little to activate adequate FRP confinement, the stiffness of the specimens was dependent on the infill column.The stress-strain curves also demonstrated that the slope of the second ascending part(when FRP confinement activated) was smaller than that of the first increasing portion for all the FRP specimens.This was attributed to the initiation of crack and the succedent crack network growth in the infill column in this stage, which resulted in a reduce in the stiffness of the infill core.The stiffness of the specimens in this stage depended on the cracked infill and the elastic modulus of the FRP.Furthermore,increased number of FRP layers resulted in greater slope in the second ascending part, indicating that higher resistance was generated for the standing support with thicker FRP liner when the axial strain was the same.This was attributed to the larger confining pressure produced in the thicker FRP liner.After reaching the peak strength, the measured stress for the unconfined plain columns plummeted due to the loss of structure integrity and the testwas immediately terminated.Likewise, at the beginning of the post-peak stage, the recorded stress of the FRP confined columns also plunged as a result of rupture of the FRP.However, the integrity of these composite columns was not fully lost and the recorded stress experienced gradual decline thereafter.

Table 2 Key results of the laboratory tests.

Fig.9 shows the influence of w/g ratio on the axial stress-axial strain behaviours of the specimens.It can be seen that, specimens with lower w/g ratio were slightly stiffer than those with relatively higher w/g ratio in both ascending portions, irrespective of the number of FRP layers.This was because the lower w/g ratio produced stronger infill column, the stiffness of the specimens depended on the stiffness of the infill in the first stage and a combination of the cracked infill and the FRP in the second stage.As indicated in Table 2,the Young’s modulus of the infill column with w/g ratio of 1.2 was 2.7 GPa, it increased to 3.1 and 4.3 GPa when the w/g ratio dropped to 1 and 0.8 respectively.

Fig.9.Typical axial stress-axial strain curves of the specimens with various w/g ratios but the same number of FRP layer.

3.5.Axial strain-hoop strain behaviour of the specimens

Fig.10 shows the typical axial strain-hoop strain curves of the specimens in each group up to the peak stress.The axial strain was calculated using the same method described in Section 3.4.The hoop strain was averaged from the readings of the survival strain gauges attached on the specimens in the circumferential direction.Similar to the observations in Yu et al.[3], some specimens experienced hoop strain gauge damage before reaching the peak stress due to the local resin breakage.As expected, the specimens confined with thicker FRP jacket generally had lower hoop strain when the axial strain was the same.This was because the thicker FRP jacket was able to provide identical confining pressure to the infill column at a relatively smaller hoop strain, compared with the thinner FRP jacket.

Fig.10.Typical axial strain-hoop strain curves of the specimens in each group.

3.6.Uniaxial compressive strength and deformability

Twenty-seven columns with and without FRP confinement were subjected to the uniaxial compressive test.As mentioned above, one specimen produced incorrect results due to eccentric loading and was omitted from this study.Table 2 presents the test results.As expected, the compressive strength of the plain unconfined column (fco) improved with decreasing w/g ratio.It can be seen from Fig.11 that, while decreasing the w/g ratio from 1.2 to 1, the UCS of the plain column increased by approximately 18%from 11.6 to 13.7 MPa.A further increase of 6.1 MPa in UCS was achieved when the w/g ratio was further reduced to 0.8.This was because the porosity of the mixture increased as the w/c ratio went up[29,30],and the compressive strength of the mixture was adversely affected by the porosity [31,32].Greater material strength may be possible if plasticisers are used to minimise the w/g ratio further.

Fig.11.Compressive strength of the specimens.

While confined with the FRP jacket, all plain infill columns experienced improvement in strength irrespective of their w/g ratios.Specifically, the strengths of the plain infill column with w/g ratios of 0.8, 1 and 1.2 increased by 90%, 150% and 130%respectively when confined by 2 layers of FRP.This was attributed to the confining pressure provided by the FRP.The confining pressure can be calculated using the following equation [33]:

where fl,ais the elastic modulus of the FRP; εh,rupthe hoop rupture strain of the FRP and D the diameter of the infill column.As shown in the above equation, the confining pressure is positively proportional to the FRP thickness, indicating that greater strength improvement would be achieved with thicker FRP liner.It is shown in Fig.12 that the strength enhancement ratio (fcc/fco) increased with increasing FRP layers.The strength enhancement ratio refers to the ratio between the UCS of the FRP confined column (fcc) and that of unconfined column(fco).For example,the strength enhancement ratio increased from 1.9 to 2.9 when the confinement FRP layers increased from 2 to 4 for the infill column with w/g ratio of 0.8.

Fig.12.Strength enhancement ratios(fcc/fco)and strain enhancement ratios(εcc/εco)of the tested specimens.

After peak stress, the FRP confined infill columns had shown various levels of residual strength (Fig.8).This was due to the occurrence of localized FRP rupture rather than complete FRP rupture after peak stress was reached.As such the FRP confined column was still able to preserve part of its structural integrity to provide confining resistance after failure.Overall, greater residual strength was found in the columns with thicker FRP and the residual strength decreased as the deformation increased.For instance,the initial residual strength of R1F4 specimens was approximately 15 MPa, whereas that of R1F2 specimens was about 9 MPa.

Due to abutment loads and caving, high load and deformation are often encountered in the longwall tailgate of underground coal mines[34].A component of the deformation in the tailgate,which normally occurs due to elastic dilation of the rock mass before failure,is considered as uncontrollable[11,35].Thus,in addition to the strength capacity, the deformability of the standing support plays an important role in roof support in underground coal mines.This is because the standing support with inadequate deformation ability tends to fail prematurely so as to compromise its rock support performance.As shown in Table 2,the axial strain at peak stress of all infill columns was significantly enhanced when confined with the FRP.Specifically, when confined with 2 layers of FRP, the axial strain at peak stress increased by 6.6, 12.0 and 18.1 times for the infill columns with w/g ratios of 0.8,1 and 1.2.The strain enhancement ratio (εcc/εco) illustrated in Fig.12 and Table 2 indicate that further deformability improvement was also obtained when thicker FRP was applied.The strain enhancement ratio refers to the ratio between the strain at peak stress of FRP confined columns ratio(εcc)and that of the unconfined column ratio(εco).The highest strain achieved for the FRP confined columns reached up to 11.8%.

3.7.Actual confinement ratio

The actual confinement ratio(fl,a/fco)refers to the ratio between the confining pressure and the compressive strength of unconfined column.Table 2 lists the confinement ratios of the specimens.Note that some of the specimens experienced hoop strain gauges failure before the peak stress was achieved, the hoop rupture strain of these specimens was assumed as the same as that of a specimen in the same group.It is clear that for the specimens with the same w/g ratio, the confinement ratio increased as a result of growing FRP thickness.This was due to the increased confining pressure provided by the thicker FRP liner.Specifically, the confinement ratio of specimens with w/g ratio of 0.8 and 2 layers FRP confinement was 0.955,it increased to 1.934 when 4 layers of FRP confinement was applied.Similar tendencies were also found in the specimens with w/g ratios of 1 and 1.2.Figs.13 and 14 illustrate that the strength and strain enhancement ratio of the specimens was generally positive proportional to the confinement ratio.

4.Proposed models for predicting peak strength and the corresponding strain

FRP has been successfully used to confine concrete structures in civil engineering for a long time.Numerous studies have been conducted to investigate the compressive behaviour of FRP confined concrete columns.Various analysis-oriented and design-oriented models for FRP externally confined concrete column have been proposed[36],whereas models for the innovative standing support described in the present study have been limited.This section presents a new design-oriented model to predict the peak stress and corresponding strain of the innovative standing support.The model includes closed-form equations that are derived from the laboratory test data in this study.

Many existing strength models adopt the following correlation to predict the strength of FRP confined concrete columns:

where k is the confinement effectiveness coefficient.In this case,the strength enhancement ratio is linearly proportional to the actual confinement ratio.The k is assumed to be various values in these models, such as 2.2, 3.3, 3.57 and 4.2 in the studies of Benzaid et al.[37],Lam and Teng[7],Realfonzo and Napoli[38]and Tamuzs et al.[39] respectively.Similar models that related the strength enhancement ratio to actual confinement ratio were also proposed by Al-Tersawy et al.[40] and Binici [41], with their expressions described in Eqs.(4) and (5) respectively.

Teng et al.[42]proposed a refined model in which the strength enhancement ratio is a function of the confinement stiffness ratio ρkand the strain ratio ρε:

where t is the thickness of the FRP jacket;Efrpthe elastic modulus of the FRP in the hoop direction; D the diameter of the confined concrete column; and εh,rupthe hoop rupture strain of the FRP.

Fig.13 illustrates performance of some of the above-mentioned models in predicting the strength enhancement ratio of the specimens in the present study.It is clear that results from these models are not in close agreement with those from the experiments.One of the possible reasons is that the parametric range in the present study does not match the ranges of these models.Since most of the existing models are derived from the laboratory test database, the accuracy of these models is affected by the parametric range of these tests [36].The maximum compressive strength of the infill material for the standing support is 19.8 MPa,which is lower than that of most concrete columns in the existing models.The concrete with greater strength is normally more brittle and its failure model tends to be more-like localized macrocrack instead of heterogeneous microcrack in low strength concrete [43].Another possible reason is that, instead of the natural gravel and sand which are usually used to cast the concrete, local Illawarra coal rejects and metalliferous tailings were used as the coarse and fine aggregates respectively to cast the infill column for the FRP standing support.As such, a new simple equation is proposed to predict the peak strength of the innovative FRP confined infill material based on regression analysis of the test data in this study:

It can be seen from Fig.13 that the proposed model is able to predict the peak strength of the innovative standing support with a high degree of accuracy.

Like the strength models, the strain enhancement ratio of FRP confined concrete is correlated to the actual confinement ratio in some of the existing strain models, such as the models of Al-Tersawy et al.[40] and Benzaid et al.[37] which are expressed in Eqs.(11) and (12) respectively.

In the model of Lam and Teng [7], a correlation between the strain enhancement ratio and the actual confinement ratio together with the strain ratio was proposed.Teng et al.[42]further refined this model by substituting the actual confinement ratio for the confinement stiffness ratio.Expressions for these two models are shown in Eqs.(13) and (14) respectively.

The performance of these strain models in predicting the strain enhancement ratio of the innovative standing support is shown in Fig.14.Compared with the other two models,both models of Teng et al.[42] and Benzaid et al.[37] had closer agreement with the test results.A possible reason for the inaccuracy of the models has been explained in the above strength models.Another possible contributing factor is the difference in strain measurement in these studies.For instance, while the axial strain was reported as the average readings of 2 LVDTs covering the 120 mm height in the middle of the concrete column in some of the existing models,the axial strain in the present study was based on 2 LVDTs measurements of the whole 300 mm height of the column.To better evaluate the strain characteristic of this standing support, a new simple strain model is also proposed as follows:

Fig.14.Performance of the proposed strain model and some existing strain models.

In order to make a quantified comparison between the proposed model and the existing models, the mean square error(MSE) and the average absolute error (AAE) were calculated using the following equations [36]:

where mod and exp are the results from the model and the experiment respectively; and N the number of tested specimens.Table 3 presents the MSE and AAE values of the models.It is clear that the proposed model had smaller MSE and AAE than the other models,which indicates the proposed model appears better than the other models in both strength enhancement ratio and strain enhancement ratio prediction for the standing supports described in the present study.

5.Conclusions

The present study is an investigation on the uniaxial compressive performance of an innovative standing support that is structurally FRP jacket confined infill column made of coal rejects,tailings and cementitious grout.The standing support is supposed to provide rock support in underground spaces such as underground coal mines.27 cylinders with and without external FRP confinement were prepared and subjected to the uniaxial compressive strength test.Based on the test results, the following conclusions can be drawn:

(1) The compressive strength of the infill column is inversely affected by the w/g ratio, whereas the workability of the mixture of the infill materials improved as a result of increasing w/g ratio.

Table 3 MSE and AAE values of the models.

(2) The standing support experienced significant improvement in both the strength and deformability when confined by the FRP jacket.More layers of FRP jacket achieved greater enhancement in strength and strain.The highest strength and strain improvement (enhancement ratio) reached up to 3.6 and 27 times the values of the unconfined samples.

(3) Various levels of residual strength were observed in the FRP confined infill columns, and the columns confined with thicker FRP generally had greater initial residual strength than those confined with thinner FRP.

(4) The innovative standing support exhibited strain-hardening behaviour with the highest axial strain at peak stress of 11.4%.The strain at peak stress of this innovative standing support appears higher than many other conventional pumpable standing supports.The standing support with large deformable ability is desirable for underground strata control as it is able to provide continuous support to the roof rock over a large roof deformation range.This helps to minimise the mine roof softening,allowing the rock mass to preserve its self-supporting capacity for as long as possible.

(5) A correlation between the compressive strength of the infill column and the ultrasonic pulse velocity(UPV)has also been established in this study.This non-destructive test is beneficial as it allows the prediction of the infill material strength without damaging the specimen.

(6) A design-oriented model was also proposed based on regression analysis of the test results presented in this study.The MSE and AAE values indicated that this model is able to make closer approximation on the peak strength and the corresponding strain of the innovative standing support than some other existing models.

Successful application of this FRP standing support is able to not only bring economic benefits but also help reduce greenhouse gas emissions.The proposed models can be used as guidelines to predict the strength and deformability of analogous FRP standing support products.Further research may be conducted to include more types of waste material into the infill column and investigate the influences of raw materials on the performance of the FRP standing support.