Ground control research continues to be a major part of the portfolio reviewed by SHMRAB. The majority of the work was carried out through ECSC and RFCS collaborative projects by Rock Mechanics Technology Ltd (RMT), UK Coal Mining Ltd and the University of Nottingham. Progress made during 2004 is reported below.
Improved roadway and face end support techniques.
This project, ECSC PR115, was completed in October 2004. RMT measured the properties of cribbing systems currently used in the UK and also investigated systems used elsewhere in the world, particularly in South Africa. The results from the laboratory test programme showed that:
Data from the tests was compared with predictions from the STOP computer program, developed by NIOSH, and with a separate mathematical model predicting Link-n-Lock performance. Comparison of results with predictions for softwood and hardwood cribs showed reasonably close correlation, indicating that input of engineering properties of European sourced materials into the STOP program would yield reasonably reliable results. Comparison of Link-n-Lock results with predictions from the separately available mathematical model showed good agreement for the 610 and 1065mm systems, but not for the 910mm system.
The STOP routine was updated to take account of European system performance in two ways. Firstly, measured mechanical properties of materials used for European cribs were stored in the STOP database for use in calculation of predicted load vs. displacement characteristics. Secondly, Link-n-Lock test characteristics were entered into the database as digitised data points from averaged test characteristics. Details of European system construction and design details, support costs, etc., were also edited into the STOP routine and the updated program placed on CD for distribution to European mining engineers. The Link-n-Lock mathematical model was placed separately onto the CD.
Also included on the CD are details of tests carried out in South Africa on a comprehensive range of support systems available there, and which may become available in Europe. These include timber crib systems, composites of timber and cement, and pumped cement packs.
The in-situ performance of Link-n-lock and cement block cribs was monitored at a UK deep mine. The instrumentation was pre-calibrated in the laboratory by installation in crib components within a compression testing machine. The monitoring quantified asymmetric loading of cribs due to poor packing to the roof and in particular point loading of the cement block cribs induced by the fitting of steel bars between the cribs which led to premature failure in some cases.
Numerical modelling was used to assess steel and mixed support systems for UK deep mines. Runs with a cambered profile roadway and support gave inferior results to those for an arched profile and support. With the addition of a full reinforcement pattern, the cambered profile and support performed better than the arched profile.
UK Coal's contribution to this project was focussed on extending the current boundaries of rock bolting reinforcement. Developing reinforcement support systems for face lines, laterals and high stressed roadway conditions.
Improved roadway drivage and ground control under high stress conditions
RMT's work on ECSC Project PR132, due for completion in June 2005, is concentrating on laboratory testing of reinforcement consumables, particularly flexible systems, with a view to revision of BS7861:2. A sub-committee of the Rockbolt Research Liaison Committee, chaired by HSE, is preparing a draft revision of the Standard, which currently only covers birdcaged cable bolts. The aim is to develop a Standard that will apply to the range of flexible tendons currently in use in the UK industry, including flexible bolts spun into resin, bird caged and other cable bolts that use cementitious grouts, and tensionable systems such as Megastrand.
A Laboratory Short Encapsulation Pull Test (LSEPT) has now been developed for flexible tendons encapsulated in resin, in which their axial bond strength and bond stiffness load transfer properties are measured in a low strength sandstone. An embedment length of 450mm was found to be suitable and tests were completed on Osborn Reflex bolts, MMTT 19 wire flexible bolts (2 types), Bridon 19 wire strand and Megastrand. Initial recommendations have been made concerning acceptance criteria for this test (300kN and 150kN/mm respectively).
Initial LSEPTs have been carried out on long tendon cable bolts encapsulated with cementitious grout. The Osborn mini cage has been selected as the benchmark tendon, as it is currently the most widely used of this tendon type in the industry, but the properties of the original double birdcage cable bolt that is no longer used will also be referred to. Further work is still required to determine the appropriate embedment length, end termination and means of confinement for this test.
It is envisaged that the load transfer characteristics of tensionable systems, such as the Megastrand, which are partly resin encapsulated and partly cementitious grout encapsulated, will be assessed against both criteria. However, additional criteria will also need to be agreed, particularly concerning the application of tension and post grouting facilities and procedures. This has been recognised as being particularly important following several instances of in-situ failure of these systems during the year.
A series of double embedment tests was carried out on Megastrand with a range of PUR products as encapsulant following interest in using PUR for combined strata injection and long tendon grouting. Tests with three different products were undertaken according to BS7861 Parts 1 and 2 (250mm and 900mm embedment respectively). None of these tests met the requirements of the appropriate parts of the current Standard in terms of bond strength, maximum load or bond stiffness. Whilst it appears possible to develop PUR products with reasonably high compressive strengths, a PUR product with stiffness approaching that of polyester or cementitious grouts for these applications has not yet been seen.
Tests were also completed on a hydraulic pre-tensioning rockbolt plate developed in South Africa. These included pretensioning strain gauged rockbolts underground to examine the load profile developed in the bolt and its variation with time. They showed that significant pre-tensioning bolt loads can be achieved with this system (200 kN was demonstrated), but they remain to some extent dependent on pressure being maintained within the plate. There is evidence that pressure loss occurs over time and this will need to be addressed. Health and safety issues centre on identifying the failure pressure of the system and the mode of any such failure. Consideration should be given to devising a means of solidifying the plate post- pressurisation, which may fulfil the aims of providing a low hazard system.
The initial objective of the University of Nottingham's input to ECSC project PR 132 was to investigate the use of pre-tensioned bolts and cables in a coal mining environment and to investigate the beneficial effects of the use of pre-stressing supports on the immediate strata surrounding a coal mine excavation. In 2004, researchers widened their objectives to consider the total environment of the rock bolt with particular attention being given to a study of the way in which hole roughness and hole annulus size affect bolt support efficiency.
The mechanism of rockbolting is controlled by a number of parameters that be categorized into three groups:
A combination of these parameters determines the performance of the rockbolt system. Failure of the bolt as a support element, should it occur, will be in one of the following parts of the system:
Every failure zone is controlled by many parameters. Therefore understanding the interaction and the sensitivity of these parameters will help the designer to develop a better design methodology.
In soft rock such as that associated with the coal measures, the rock/grout interface is thought to be the dominant potential failure zone that controls the failure of the rockbolt. Previous research has shown that full-column resin-grouted bolts rely on the mechanical interlock of the resin interstices within the borehole wall and along the surfaces of the reinforcement rod to secure the surfaces together. Accordingly, the peak shear strength at the interface can reduce by up to 30% when the interface changes from rough to smooth. Furthermore, previous published work has shown that random grooving of a borehole increased anchorage performance three-fold when compared to a smooth clean borehole surface. Therefore it has been decided to address the effect of the borehole roughness on the performance of the rockbolt in more detail in the laboratory.
The direct shear test is a well-established test whereby the shear strength parameters of the materials can be measured. The test method involves maintaining a fixed normal load and introducing shear loading by relative movement of the two halves of a shear box. After determining the maximum shear strength for at least two samples at different normal load, values for cohesion and friction angle can be derived. Experiments have been conducted on the rock/rock and rock/grout interfaces with differing levels of surface roughness and grout thickness to further quantify these factors.
Triaxial tests have been conducted to validate the test results. The results show that the shear box tests underestimate the shear parameters of both types of rock by almost 50% when compared with the results obtained from the triaxial test. This inconsistency between the two results can be attributed to two reasons. Firstly, the axis of reaction and action in the shear box are slightly mis-aligned leading to a moment arm which cannot be avoided completely. This in turn develops additional stresses at the shear plane. Secondly, the shear stresses are not evenly distributed over the shear plane and that causes a shear stress concentration over half of the shear plane. This work continues.
HSE contributed to the funding of RMT's work on projects ECSC PR132 through a research contract entitled 'Testing and standards for rock reinforcement consumables'. It also supported the review of BS7861:2 by commissioning HSL to investigate samples of failed rockbolts and flexible reinforcement from two collieries where there had been falls of ground associated with disturbed areas of strata.
The purpose of the investigations were to:
The samples were visually examined, measured and photographed. Material was removed from the samples for mechanical testing, chemical analysis and microscopic examination. The results of the tests confirmed that the rockbolts and flexible reinforcement complied with the relevant standards.
The findings from both investigations indicated that the failures were caused by overloading, i.e. the rockbolts and flexible reinforcement at both installations had been subject to forces in excess of their load carrying capability. In most cases, the forces imposed on the strata reinforcement had involved a significant lateral component, which indicated that failure of the rockbolts was associated with significant shearing displacement of the strata. There was some corrosion of the rockbolts from one of the collieries but there was no evidence to indicate that this had caused any significant weakening prior to the failures.
Improved rock stress measurement and analysis for planning underground coal mines
RMT is project co-ordinator for ECSC project PR135 due for completion in June 2005. Their work during the year has been to continue to investigate the feasibility of determining in-situ rock stress from analysis of acoustic emissions during laboratory testing of rock samples recovered from the ground (the Kaiser effect).
Acoustic emission tests were undertaken on core recovered from in-situ stress measurements by overcoring at two sites in a French coal mine, from an overcoring stress measurement at Tara mine in Eire, from one successful and one unsuccessful overcoring stress measurement site at a South African platinum mine and on core from two boreholes from a shallow South African coal mine. Also during the year, improvements to the laboratory testing regime have been investigated based on recommendations in a draft ISRM Suggested Method.
The results from the French coal mine were variable but when summarised and compared with the stresses measured by overcoring it was evident that there was some correlation with the more reliable Kaiser effect values. This was an exacting test of the method as the stress regime in this case was near isotropic, considered the theoretically least favourable situation. The results from Tara mine gave superficially good correlation but it was concluded that the interpretation could have been heavily biased by a pre-knowledge of the overcoring result. The Kaiser effect result for the South African platinum mine agreed closely with the successful overcore result and the Kaiser effect result for the unsuccessful overcore site agreed closely with the other site. The results from the shallow South African coal mine were less conclusive but did give an indication that the maximum horizontal stresses were within the expected range and probably in the expected direction.
Currently the technique still appears to have considerable promise for characterisation of the in-situ stress field when used in conjunction with selective overcoring. However its applicability to a site where there is no pre-knowledge of the expected stress regime is still questionable and requires further investigation. A database of RMT's Kaiser effect results to date has been set up and this now contains 176 data points. This will be analysed during the closing months of the Project. Tests will also be undertaken on man made engineering materials such as brick and concrete and further "blind" tests undertaken.
The work conducted at the University of Nottingham for this project has been divided into two areas. The first of these, a novel treatment of uncertainty in geotechnical modelling, was reported in last year's SHRMAB review. The other area, the variability of the key rock parameter of stiffness with imposed load, was addressed in 2004.
The phenomenon of the dependency of the elastic modulus on the level of stresses has been observed in a number of studies conducted on both intact and jointed rock. With the benefit of the published literature and the internally available data at the University of Nottingham, the phenomenon of stress-dependent stiffness has been reviewed. The work has taken the form of further experimental investigations of this phenomenon on a wide variety of fractured rocks.
Values gathered from triaxial test data imply that confining stress influences the stiffness of rock. This has its greatest effect at low confining stress (generally below 10Mpa) although there is a general monotonic increase in the Young's modulus up to the maximum confining stress applied (29MPa). With the increasing stress, there is a tendency for the deformation modulus to level out. The main reason for this increase is most likely due to the closure of joints and micro fractures that existed previously in the samples. Moreover, it has been noted that these increases of the Young's modulus are significant and in some cases can be as much as 20-43 times of the initial value (i.e. at zero confinement). Testing results are continuing to be processed seeking a more usable and robust equation that predicts stiffness of intact and broken rock under confinement.
Geotechnical modelling, classification and exploration for safe and efficient mine layout and tunnel support design
RMT is also project co-ordinator for RFCS project CR 03011, due to run to August 2006. Their main objective under this Project is to improve the ability to employ numerical modelling to solve mine layout and support design problems in UK coal mines and thus avoid falls of ground. The main modelling codes being used are FLAC, FLAC 3D and MAP3D. Some use is also being made of STAAD.Pro for modelling steel support structures.
Work on mine layout taking account of multi-seam interaction has allowed quantification of two important issues which have long been suspected:
Procedures to interpret modelling analyses results are being developed to highlight areas of potentially severe interaction and to enable the interaction risks associated with alternative layouts to be compared. These techniques are being trialled to assist in planning the layout of future workings at two UK Coal mine sites.
Work has commenced on developing the use of FLAC3D to model a short representative length of roadway. This will enable improved modelling and hence design of support systems, particularly for rock-bolting patterns. For 2D models, bolts are "averaged" along the length of the roadway so restricting the extent to which different bolting patterns can be represented and properly compared.
Site specific plots have been produced clearly showing large movements in the ribsides with bulging of the coal between the rib bolts along the length of the roadway. This allows the effect of varying the bolt spacing along the roadway to be better investigated. Rib support has generally received less attention than roof support, but has caused serious problems at a number of sites. An investigation of rib support was prompted by the experience at one particular site, where a change in rib behaviour was observed associated with a change in the drivage horizon.
Pull tests on rib-bolts at the site demonstrated that the bond strengths being achieved were low. 28 out of 35 bolts tested pulled out at a load of less than 50kN, perhaps due to deterioration over time or loss of resin into joints and fractures. The modelling demonstrated that rib bolts were much less effective at bond strengths below 50kN/m. The main alternatives considered to improve the rib support were, the use of rib cables, rib injection or the provision of posts along the ribside. Computer modelling is being used to examine the effectiveness of the first two of these, with particular regard to the performance during face retreat.
Computer modelling of strata-injection as a support technique is novel. The failed coal in the ribside will have lost strength. It is presumed that rib injection will increase the strength of the failed coal whilst the existing state of stress remains unchanged. If the ribside is subsequently loaded further during face retreat, the stress in the injected coal will increase until it exceeds the new value for the strength at which point it will again start to fail.
Model results showed rib injection to be very effective at reducing the rib displacements. However, they also showed that the injected rib could fail again, ahead of the retreating face. Once the coal has again failed, the ribside is in a similar condition to that before injection and, in the absence of effective bolted support or other restraint, has the potential to slump or topple out.
In contrast to the rib injection, rib cables were less effective in reducing the rib displacements, but would remain intact. For the cable bond strengths thought likely to be achieved, the tensile loads developed in the cables remained well below the tensile strength of these tendons.
The University of Nottingham's contribution to this project in 2004 has investigated the effect of time on the mechanical behaviour of selected coal measure rocks. The literature review related to these properties has been completed successfully but experimental work continues. Characterisation of these properties for modelling purposes is being conducted concurrently with the experimental work.
In common with any rock mass, coal measure response may be time-independent or time-dependent and the material may behave elastically or may yield according to the confining pressure and the applied stress level. From literature, there are essentially two classes of time-dependent behaviour observed in the laboratory:
In comparison with time-dependent strength, time dependent deformation has been extensively studied under the general title of 'creep'. In coal mines, it is generally accepted that after the initial excavation of an entry, most coal-measure rocks deform over time, even when no mining is taking place nearby. This deformation is most evident in the form of roof sag, floor heave, pillar dilation, or shearing along bedding planes and joints. Three types of tests have designed and are being conducted to investigate this phenomenon:
Test-A using conventional creep testing procedures to investigate the effect of time on intact samples of a coal measure rock under a constantly maintained uniaxial stress condition.
Test-B using multi stage triaxial compression testing (MSTT) procedures to further study the effect of confining pressure on stiffness of a number of intact and fractured samples obtained from a range of different rocks (sandstone, mudstone, siltstone and granite). (see PR 135 above)
Test-C using MSTT procedures to quantify the effect of strain rate (time) and confining pressure on stiffness of the rock investigated in Test-A. Currently, intact rock samples are being tested; but similar testing procedures will be conducted on fractured rock samples as well.
Initial results from test A were interesting. Under relatively low constant stress (≈19% of its UCS) the rock samples showed only elastic deformation (i.e. no time-dependent behaviour). At higher constant stress, the sample experienced some continuous time-dependent strain. It is thought that this time dependent deformation might be only a primary creep and could be followed by secondary creep (with a steady state strain rate) if the stress had been left for longer.
This was supported by a single test conducted at constant stress = (≈65% of UCS). The secondary creep did not occur before approximately 1750 minutes. Therefore, the testing time was extended for at least 3000 minutes (≈ 2days) for each stress application/stage. Although these tests are time consuming, the procedure has allowed a steady state strain rate to be obtained, which is of great importance for modelling and characterising time-dependent behaviour. Upon completion of the experimental programme, the test data will be analysed and parameters that represent time dependent properties for modelling purposes will be defined.