Variation in load transfer of a fully encapsulated rockbolt

PC Hagan

The University of New South Wales (UNSW), Sydney

 

A diverse range of rockbolt designs and resin anchors are available for use in underground mines. Despite their widespread use, the factors that affect the performance of a rockbolt are not fully understood. The laboratory-scale rockbolt test facility at the UNSW Mining Research Centre is being used to gain a better understanding of the behaviour of rockbolts leading to an improvement in ground support.

A test program investigated the nature of load transfer between a fully encapsulated rockbolt and the surrounding rock mass.

It was found that with all other factors held constant, there were differences in the behaviour of load transfer depending on the method of loading. The typical pull-out arrangement as used in most field test work resulted in a different load transfer function to that observed when the load on the rockbolt was induced through bed separation.

 

 

Introduction

In recent years, fully encapsulated rockbolts (FERB) have become a key element in the design of ground support systems especially in the Australian coal mining industry. One of the main drivers for the acceptance of fully encapsulated rockbolts is that they provide “…maximum support capacity at the point where bed separation could occur. (Moreover) this mechanism could operate simultaneously at many points along the rockbolt” (Gray, Hunt and Fabjanczyk, 1998).

FERB enhance the material characteristics of rock strata by binding together adjacent strata. This increases the stiffness of the composite strata consequentially increasing its resistance to movement due to loading on the strata.

There are four elements that interact in an FERB system; the rock mass, resin (or grout), steel bolt, and the plate and nut assembly. Studies have reported on the effect of changes in one or more of these elements in terms of the overall performance of the rockbolt. But in one area there are conflicting viewpoints concerning the nature of the variation in axial load along the length of the FERB.

There are several models that deal with the variation in axial load. One model predicts the rate of load reduction is linear while another claims there is an exponential decay in load with distance along the rockbolt. The models also differ in terms of the length over which the load reduction occurs. One model assumes the load applies over the entire length of embedment while another assumes the load acts only over some proportion of the rockbolt.

The ramifications of the differences in the variation can influence the design in terms of required length of a rockbolt as well as the strength of the rockmass adjacent to bed separation capable of withstanding the high loads transferred from the rockbolt.

This paper will present the results of laboratory test work and numerical modelling that investigated the load transfer behaviour of FERB.

LOAD TRANSFER

FERB are considered a passive support system which is activated through deformation of the rock mass. Li (2000) describes two forms of deformation that can activate a fully encapsulated rockbolt these being:

·      continuous - in the case of a relatively homogenous soft rock or in extremely fractured ground; and

·      discontinuous - in the case of a blocky rock mass or where there are differences in the relative movement between strata.

Deformation of the rock mass induces a tensile axial load in the rockbolt. The tensile load acts like a spring to counteract the deformation, clamping together the rock mass as shown in Figure 1. In coal mining, deformation is thought to commonly occur between differing rock strata and deformation tends to be exacerbated when there are significant differences in the material properties between adjoining strata.

Figure 1. Load transfer between rock and rockbolt
via the grout. (after Gale, 1990)

The axial tensile load on the rockbolt is greatest at the point of bed separation and decreases with distance along the rockbolt as the load is transferred into the rock mass. This transfer is accomplished through the resin which couples together the rockbolt and rock mass. As a result of the load transfer, a shear stress is induced in the resin. This shear stress also varies with distance along the rockbolt and is equal to the rate of change in the axial load on the rockbolt. Where the rockbolt is not coupled to the rock mass via the resin, no shear can developed and no load transfer can take place.

Effectiveness of a rockbolt is often equated with its ability to transfer load to the rock mass. Serbousek and Signer (1987) defined load transfer as the change in load with respect to distance along the rockbolt. Gray, Hunt and Fabjanczyk (1998) defined load transfer in terms of the maximum stress generated per unit area of the rockbolt. More effective support systems are considered as having a higher load transfer capacity.

The rockbolt becomes ineffective either when the induced tensile load exceeds the strength of the steel rockbolt, the shear strength of the resin is breached or the load transferred into the rock mass exceeds the rock strength. In most cases of correctly installed FERB, failure results from the latter two causes.

Non-linear load transfer

The variation in load transfer was examined by Farmer (1975). He developed a model that predicted an exponential decay in load with distance along the rockbolt. Subsequent tests by Farmer using different applied loads and two anchorage lengths verified the model. There was reasonable agreement with the model up to a given load at which point it was postulated that de-bonding occurred along the resin/bolt and/or resin/rock interface.

Similar observations were made by Hawkes and Evans (1951) and Signer (1990) with a non-linear reduction in load with distance. Interestingly, Hawkes and Evans found the nature of the behaviour depended on the level of externally applied load on the rockbolt as is shown in Figure 2.

Nitzsche and Haas (1976) used a finite element approach to develop their model. The model assumed axial symmetry of the stress distribution. The model also indicated a non-linear reduction in load with distance along the rockbolt; however, they also found that most of the load was dissipated within the first 400 mm from the free surface with little confining pressure generated at the far end of the rockbolt.

Figure 2. Variation in load with distance along a fully grouted rockbolt
three levels of applied load.
(adapted by Li, 2000)

Linear load transfer

Serbousek and Signer (1987) also developed a finite element model that predicted a non-linear variation in load with distance. The model indicated the rate of load reduction was sensitive to the modulus of the resin as well as the modulus of the rock mass.

Subsequent laboratory tests revealed results that were not always in agreement with those predicted by their model. First, the rate of load reduction appeared to be linear as shown in Figure 3; at least for the majority of the load transfer. Second, the rate of load reduction was a function of the applied load on the rockbolt. Third, they concluded that length of the rockbolt had some effect on the load transfer characteristics though nearly all the load was dissipated within the first 550 mm. Finally, both grout type (resin and gypsum) and hole diameter had little effect on the rate of load transfer.

Figure 3. Load distribution for rockbolts of differing length.
(after Serbousek and Signer, 1987)

Radcliffe and Stateham (1980) undertook field studies using 50 strain-gauged bolts in bedded roof strata. They found a linear reduction in load transfer over the entire length of a rockbolt. Similar results which were also based on field studies have been reported by Fabjanczyk and Tarrant (1992).

Load variation

There would appear to be several conflicting views concerning the variation in load transfer of fully encapsulated rockbolts. One view accounts for non-linear load transfer while the alternate accounts for linear load transfer; both have been observed in the laboratory and in the field. With regard to distance, some observations have indicated load transfer is constant and independent of bolt length while for others, load transfer can even extend beyond the length of the rockbolt.

Various reasons have been postulated for these apparent differences. One being that it depends on whether the load on the rockbolt was above or below the elastic limit of the rockbolt. Another reason being due to differences in the nature of load application on the rockbolt.

Results of testwork examining the second of these factors are presented in this paper.

Numerical Modelling

A study using PHASE2, a 2-dimensional axi-symmetric, elasto-plastic finite element program (FEA), investigated two different loading methods. Whitaker (2001) postulated that the differences between the various camps might in part be due to differences in the method by which the load is applied to the rockbolt.

In a conventional rockbolt pull-out test, a hydraulic cylinder is used to apply an axial load to the rockbolt. In the process of applying the tensile load, the hydraulic cylinder reacts against the free surface of the rock, inducing a compressive load in the rock mass. This arrangement is illustrated in Figure 4.

Figure 4. Arrangement for loading a rockbolt in a pull test.
(after Galvin et al, 2001)

In the second case, a tensile load is induced as a result of bed separation with no confinement of the free surface. This arrangement is illustrated in Figure 5. The figure shows the absence of any compressive load on the rock surface. In such a case, only a small displacement is required to induce a tensile load in the rockbolt.

 

Figure 5. Application of load in a rockbolt due to bed separation
 (after Galvin et al, 2001)

Modelling laboratory testing

A modelling mesh was generated to examine the affect of loading a rockbolt in a manner described by Offner, Galvin & Fabjanczyk (2000) of the testing facility at UNSW. The mesh and dimensions of the testing cylinder are shown in Figures 6 and 7. The model accounted for differences in the material properties between the resin, steel and rock sample; the latter was based on a cementitous grout used to simulate rock.

Figure 6. Generated mesh used to model laboratory test regime.
(after Whitaker, 2001)

Hence differences in the mechanics and stress regime are likely between the two loading arrangements.

Numerical modelling by Whitaker (2001) indicated differences in the variation in load along the rockbolt as well as the stress distribution in the surrounding rock strata between the two loading arrangements.

Figure 7. Dimensions of the biaxial cell test facility.

The FEA model was used to investigate the effect of changes in applied external load on the rockbolt, confinement and dilation angle. Four levels of applied force were investigated - 300, 400, 500 and 600 MPa; three level of confinement – 0 2.5 and 5 MPa; three dilation angles – 0°, 15° and 30°.

In summary, Whitaker (2001) made the following observations:

·      displacement of the bolt increased with load reaching a displacement of 0.34 mm at 600 MPa.

·      as load increased, a greater magnitude of stress existed at any given distance into the test sample.

·      load on the rockbolt tended to vary at an exponential rate as shown in Figure 8.

Figure 8. Variation in bolt tension with distance along the rockbolt
in pull-out test arrangement.

·      zones of yielding were observed in both the test sample and resin. At a load of 300 MPa, the zone of yielding extended 100 mm into the test sample and 50 mm into the resin. At 600 MPa, the yield zone extended further into the rock mass to a depth of 190 mm as shown in Figure 9.

Figure 9. Increase in the yield zone within the test sample
with an increase in bolt tension from 330 to 600 MPa.

·      confinement of the test sample had a marked effect on both tension and yield. At a distance of 80 mm along the bolt, the tension on the bolt was 90 MPa in the case of no confinement compared to 170 MPa with confinement. In terms of the yield zone, confinement reduced the yield zone which tended to increase the stiffness of the system as shown in Figure 10. It would appear therefore that confinement can have a significant role on the effectiveness of the bolting system.

Figure 10. Confinement (left) tended to reduce the extent
of the yield zone compared to having no confinement (right).

·      changes in dilation angle had little discernible on bolt tension and the yield zone. There was only a marginal change in the stress distribution at smaller dilation angles.

Modelling parting situation

In this scenario, loading of the rockbolt was modelled on bed separation at a parting in a rock mass. A schematic of the mesh used in this analysis is shown in Figure 11.

Figure 11. Mesh used to model bed separation.

In summary, analysis of the alternate FEA model found the following.

·      displacement of the bolt increased with load on the bolt achieving a displacement of 0.61 mm at 600 MPa.

·      tension along the bolt reduced with distance from the parting in a near linear fashion. The rate of reduction being a function of the bolt load as shown in Figure 12.

Figure 12. Variation in bolt tension with distance
 along the rockbolt from point of parting separation.

·      the zone of yielding varied with bolt load. Higher loads lead to a greater extent of the yield zone as shown in Figure 13. The yield zone was conical in shape on either side of the parting. Shear failure was also evident though less extensive then the tensile yield zone.

·      confinement had a similar effect as in the previous model on displacement and yield. An increase in confinement reduced displacement as well as the extent of the yield zone.

Figure 13. Increase in yield zone with increase in bolt tension of 300 MPa (left)
and 600 MPa (right).

Laboratory Testwork

The results from the two FEA models indicated different loading arrangements are likely to have some affect on the load variation of along the rockbolt and the stress distribution within the rock mass. A test program was designed to verify these findings using the rockbolt test facility at UNSW shown in Figure 14.

 

Figure 14.  Laboratory equipment comprising hydraulic cylinder,
biaxial cell and strain gauged rockbolt.

A cementitous grout was used as the test sample in the program. The grout had been used in other testwork were consistency in material properties was required between test specimens. The grout was poured into a 145 mm diameter cylindrical polypipe mould and left to cure for 28 days. The length and diameter of the test sample were 232 mm and 140 mm respectively. The test sample was contained within a biaxial load cell and a 10 MPa confinement load applied during each test.

A borehole diameter of 26 mm was drilled in the test sample to a depth of 175 mm into which a mix-and-pour resin was added to encapsulate the rockbolt.

A strain-gauged Celtite CX grade rockbolt was used to measure the variation in load along the length of the rockbolt. The diameter of the rockbolt was 21.7 mm. Six pairs of strain gauges lay within the encapsulated section of the rockbolt. The total length of encapsulation was 168 mm.

The strain gauges were equally spaced at 30 mm intervals with the final pair 10 mm from the end of the rockbolt. A further pair of strain gauges lay outside the zone of encapsulation and was used to verify the magnitude of applied tensile load on the rockbolt. The strain-gauged rockbolt and assembled sample are shown in Figure 15. Prior to testing the strain-gauged rockbolt was calibrated in an Avery Universal test machine.

Figure 15. Strain-gauged rockbolt showing spacing of gauges (above) and assembled test sample (below).

Bolt loading

Two configurations of load application onto the instrumented rockbolt were investigated. The first configuration replicated the conventional pull-out test using a hydraulic cylinder applying the load and acting against the free surface of the test sample. This arrangement is shown in Figure 16. Six levels of force were applied ranging from 5 to 30 kN and the load variation along the rockbolt measured at each load interval.

Figure 16. Conventional loading arrangement with hydraulic cylinder
bearing against the rock’s free surface.

The second configuration was intended to replicate loading of a rockbolt due to bed separation; simulating the FEA model of parting separation by Whitaker. In this arrangement, the reactive force from the hydraulic cylinder was transferred to the opposite end of the rock sample through steel plates placed at both ends of the biaxial cell and the thick walled biaxial cell. This arrangement is illustrated schematically in Figure 17. The opposite end of the test sample was fastened to the steel plate with three studs each encapsulated in the test sample to a depth of 50 mm using the mix and pour resin – these three studs are shown to the right in Figure 15. This arrangement avoided the need to apply a load to the free surface of the test sample. Instead a tensile stress regime was created at the opposite end of the rock sample.

Figure 17. Schematic of the two loading arrangements
 pull test (above) and bed separation (below).

RESULTS & DISCUSSION

The results for the two loading configurations are summarised in Tables 1 and 2 respectively.

Table 1 indicates that load decays with distance along the rockbolt. The loads indicated at a distance of -57 mm are a measure of the load on the rockbolt. It is interesting to note that for each level of applied load there is a positive measured residual load 10 mm from the end of the rockbolt. In the case of 30 kN load, the residual load was equivalent to 6% of the external load.

Table 1

Variation in load along length of a FERB
in pull-out test configuration.

Distance into rock from free surface (mm)

-57

8

38

68

98

128

158

Nominal applied
load (kN)

Measured load along rockbolt

5

3.7

4.2

3.6

2.3

1.5

0.6

0.3

10

10.2

10.9

8.2

4.8

2.8

1.2

0.4

15

15.5

16.2

11.6

6.9

3.9

2.0

0.9

20

20.1

20.4

14.3

8.4

5.1

2.6

1.0

25

25.5

26.0

17.8

10.6

6.3

3.4

1.4

30

30.5

30.8

20.8

12.6

7.5

4.2

1.8

 

Table 2

Variation in load along length of a FERB
in bed separation configuration.

Distance into rock from
free surface (mm)

-57

8

38

68

98

128

158

Nominal applied
load (kN)

Measured load along rockbolt

5

4.9

5.3

4.1

2.5

1.2

-0.1

-0.4

10

9.9

9.9

7.0

4.1

2.0

0.4

-0.3

15

15.4

14.0

9.2

4.9

2.2

0.6

-0.1

20

20.4

17.7

11.8

6.4

3.0

0.9

0.0

25

25.5

22.3

16.0

9.5

5.1

2.0

0.3

30

29.8

26.5

19.4

12.1

6.8

3.1

0.7

 

In the case of the bed separation configuration, the results shown in Table 2 indicate a much lower residual load. In fact a compressive load is indicated. This may be due to the method of measurement. The load at each mark along the rockbolt was measured manually hence there may have been some relaxation by the time the measurement was made at the final mark. To overcome this, measurements will be made with a data-logger meaning all readings along the bolt will be made almost simultaneously.

The values from Tables 1 and 2 are graphed in Figures 18 and 19.

Figure 18. Variation in load along the rockbolt in
the conventional pull-out test configuration.

Figure 19. Variation in load along the rockbolt in
the bed separation configuration.

The two graphs illustrate the variation in load as measured using the strain gauges. The load within the zone of encapsulation is represented by the solid points in the graph; that is at distances ranging from 8 to 159 mm. The six outlined points in the graphs represent each of the loads measured at the -57 mm mark. The greyed in-filled box represents the length of the test sample of 232 mm. The darker grey rectangle along the x-axis represents the rockbolt with an encapsulation length of 168 mm.

Over the range of loads applied, the nature of the load variation appeared to be similar. There was a non-linear variation with distance as load was transferred from the rockbolt into the test sample. The rate of load transfer was not constant but increased with the level of applied load.

While the overall nature of the variation appeared similar there were some distinct differences in observed behaviour.

First, in the case of the pull-out test arrangement there was, for each of the six levels of applied load, a slight increase in the load within the test sample as indicated by the difference in the load measurements at the -57 mm and 8 mm marks respectively. This suggests confinement of the free surface by the hydraulic cylinder altered the stress field around the rockbolt. It is likely that a combination of compressive and shear stresses were generated as a result of the confinement by the hydraulic cylinder.

In the alternate loading arrangement, it appeared that load transfer from the rockbolt into the rock sample commenced almost immediately as indicated by the reduction between the -57 mm and 8 mm marks.

The second difference in behaviour concerned the level of residual tensile load indicated at the end of the rockbolt. In the case of the pull-test arrangement, a residual load was observed at the end of the rockbolt. The level of residual load increased with applied load.

In the alternate arrangement, no residual load was evident for all levels of applied load. Load transfer occurred over the full length of encapsulation.

CONCLUSION

Two loading arrangements of a fully encapsulated rockbolt were investigated to determine the affect if any on the resultant load transfer behaviour. The two loadings arrangements examined were the conventional rockbolt pull-out test and bed separation.

The arrangements were analysed using a finite element model and a physical model in the laboratory.

It was found that load transfer occurred over the full length of encapsulation independent of the level of force applied and method of application. Previous research indicated that load transfer could be limited to less than the full length of the rockbolt. The physical model was limited to an encapsulation length of 168 mm being equivalent to about 7.7 rockbolt diameters. There may have been a scaling effect present given the short length of encapsulation relative to the rockbolt diameter. A smaller diameter rockbolt or longer length of encapsulation needs to be examined to determine whether similar behaviour can be replicated.

There was some commonality as well as differences in the observed nature of the load transfer between the two loading arrangements. Significantly, the load transfer was non-linear with 50% of the load transfer taking place within approximately 54 mm  or 2.5 rockbolt diameters from the free surface in both arrangements.

It should be noted that the level of load actually applied was far less than that required for the bolt to yield which in similar arrangements has been measured at approximately 250 kN. It is intended to examine the behaviour at higher levels of applied load up to the yield point over a longer length of encapsulation in a longer biaxial cell. These changes may provide some clarity as to the differences indicated between the finite element model and physical model.

ACKNOWLEDGEMENTS

The author wishes to thank the Australian Coal Association Research Council (ACARP) for funding the project and to Celtite Pty Ltd for providing test materials. The author also acknowledges the contributions made by Alison Whitaker, Glenn Dawson and Jon Balcomb to the rockbolting research program at the UNSW Mining Research Centre.

REFERENCES

Fabjanczyk, M and Tarrant, G C, 1992. Load transfer mechanisms in reinforcing tendons, in Proceedings 11th International Conference on Ground Control in Mining (The Australasian Institute of Mining and Metallurgy: Melbourne), pp 1-8.

Farmer, I W, 1975. Stress distribution around resin-grouted rock anchor. International Journal of Rock Mechanics, Mining Science and Geomechanics Abstracts, 12:347-351.

Gale, W, 1990. Role of Rock Bolting in Coal Strata Control. Training Program, (Strata Control Technology: Wollongong).

Galvin, J M, Offner, J C, Whitaker, A, Fabjanczyk, M and Watson, J O, 2001. Establishing Anchorage and Failure Mechanisms of Fully Encapsulated Roof Support Systems. ACARP Project C7018, UMRC Research Report RR2/01, (UNSW Mining Research Centre: Sydney) ISBN 0733417590.

Gray, P, Hunt, N and Fabjanczyk, M W, 1998. New developments in ground support with particular reference to high capacity, high load transfer rock bolts, in Proceedings International Conference on Geomechanics/Ground Control in Mining and Underground Construction, (University of Wollongong), pp 513-524.

Hawkes, J M and Evans, R H, 1951. Bond stresses in reinforced concrete columns and beams. Journal of the Institute of Structural Engineers, 24(10):323-327.

Li C, 2000. The load-bearing process of fully coupled rock bolts, in Proceedings MassMine 2000, (The Australasian Institute of Mining and Metallurgy: Melbourne), pp 933-935.

Nitzche, R N and Haas, C J, 1976. Installation induced stresses for roof bolts. International Journal of Rock Mechanics, Mining Science and Geomechanics Abstracts, 13:17-24.

Offner, C, Galvin J and Fabjanczyk, M, 2000. Evaluating anchorage mechanisms of fully encapsulated rock bolts, in Proceedings 19th Conference on Ground Control in Mining, Morgantown, WV, August, pp 255-260.

Radcliffe, D E and Stateham, R E, 1980. Stress Distribution Around Resin Grouted Bolts, USBM RI8440, (USA Dept of Interior).

Serbousek, M O and Signer, S P, 1987. Linear Load-Transfer Mechanics of Fully Grouted Roof Bolts. USBM RI9135 (USA Dept of Interior).

Signer, S P, 1990. Field Verification of Load Transfer Mechanics of Fully Grouted Roof Bolts. USBM RI9301 (USA Dept of Interior).

Whitaker, A, 2001. Critical Assessment of Past Research into Rock Bolt Anchorage Mechanisms. UMRC Research Report RR3/01, (UNSW Mining Research Centre: Sydney) ISBN 0733417590.

 

Original version of manuscript published in Proceedings 23rd International Conference on Ground Control in Mining, Morgantown, WV, August 2004.