GRANIT is a non-destructive anchor integrity testing system used for quality control, exception testing, load and effective free length diagnosis, bond verification and troubleshooting of anchors from 10mm– 75mm diameter and lengths up to 40m(1). It has been developed by the UK’s Universities of Aberdeen and Bradford over the last ten years, from an instrumented bolt programme in two road tunnels in North Wales. For the last 5 years, Amec has provided the field work and practical development and as sole licensee now offers the system for commercial use.

The system works by attaching a purpose built pneumatic impact device to the head of an installed anchor. This device applies a small impulse to the anchor. An accelerometer and data acquisition system then captures the vibration response signal which is then processed and fed into a neural network previously “trained” to recognise characteristics of the signal and interpret these to provide diagnosis of load and free length. No special preparation of the bolt and no coupling medium are required. The device is simply clamped on where screw threads are not present.

Need for integrity testing

Regulatory authorities frequently demand that managements fulfil their safety obligations regarding the application of anchor systems by regular testing and provision of hard data to back up safety claims(2). It is important that there is a method of confirming that the specified bolt/anchor installation quality is being achieved and that the bolts are performing to specification. There is no visual way of detecting bolt system out-spec installation or incipient malfunction except when visible movement occurs in the strata above or actual deformation of the bolted roadway. Prior warning would be advantageous as remedial measures can be taken, enhancing safety whilst avoiding interruptions to production.

As an example of the need for adequate anchor encapsulation and bonding, a recent survey of roof falls in deep South African minesâ showed that over 70% were less than 0.5m thick. This shows how critical full bolt bonding and/or pre-tensioning is, to ensure that such shallow falls are minimised. In particular, poor encapsulation/bond at the proximal end of the bolt is likely to predispose the roof to this type of shallow fall especially where little or no pre-tension has been applied to the bolt.

Laboratory testing

GRANIT has been developed to address these problems by enabling assessment of unbonded length and load under the head of the rockbolt. This development is backed up by extensive laboratory testing and computer modelling of unbonded/free length(4) and load diagnosis(5). Laboratory testing utilised full scale anchors, bonded in concrete in steel gun-barrels, of various sizes and lengths and with a wide variety of anchor head hardware.

A computer model has been developed at the University of Aberdeen that enables accurate mathematical modelling of a variety of anchorage types. It can predict the likely responses of anchorages to GRANIT testing and to verify real field data.

The model can also generate data for the initial training of neural networks, without extensive, time consuming collection of field data. This mathematical model is a very important and useful part of the overall GRANIT system, and is now being further refined

Proving trials

Prior to full scale application, the system was used in surface trials at Amec’s test site in Swynnerton, UK. These trials were designed to replicate as closely as possible the conditions found in applications, such as mines, that use resin bonded bolting systems. Further extensive trials were also undertaken at another site at Kibblestone, UK, where the anchors were designed to replicate single strand anchors often found in tunnels.

The aim of the trials was to establish the accuracy to which GRANIT could estimate unbonded length and load under the head of the bolt and to verify exception testing. The three results: Actual Unbonded Lengths (mm) of 280, 110, 300 corresponding respectively to Diagnosed Unbonded Lengths (mm) of 285, 110, 285, are typical of the accuracy obtained by GRANIT in tests to diagnose the unbonded length of the bolt at the Swynnerton test site. The bolts were installed in a manner as close as possible to typical underground construction site methodology. Many thousands of load diagnosis tests were carried out on rock bolts and other anchorages installed in the test site at Swynnerton and at the natural sandstone site at Kibblestone.

Figure 2 shows the diagnosis of load in one test bolt. This programme gave the confidence that the system could diagnose the unbonded lengths of the bolts in the roadway of coal mines and the load under the head of these bolts

Having proved that the GRANIT system could accurately determine unbonded length and load, the system was tested underground at UK Coal’s Thoresby Colliery, in a Longwall main access roadway and panel main gate. Tests were undertaken in two sectors of the mine and at different distances along the roadways in the two sectors. In one sector the bolts had been installed about two months prior to testing and in the other about 30 months prior to testing. The aim was to see if there was any substantive difference between the two regions with different ages of bolts. Data was taken to allow diagnosis of unbonded length and load.

Unbonded length

The results of the tests to ascertain unbonded length, showed that there was a large distribution of values of unbonded lengths, which was inconsistent with the design specification required for coal mining practice(2). In this region of the mine, there is a wide variation in unbonded lengths (Figure 3). In fact, two of the bolts have unbonded lengths of around 500mm. The reason for these large unbonded lengths may be due to fissuring in the rock resulting in leakage of the resin into the fissures during installation, resin loss at the mouth of the hole during or shortly after installation or simply oversize drilling causing short-run of the displaced resin down the hole.

This is a very important result for two reasons. Firstly it shows that, although these bolts visually met the design criterion for total/full length encapsulation, full bonding had not been achieved over the full length. This is important for the prevention of shallow roof falls. Secondly it shows that GRANIT can be used as an effective tool for the monitoring of the installation practice of the rockbolts. The use of the GRANIT system in this way will enable bolts that are out of specification to be identified and any necessary remedial action taken.

Difference in unbonded length distributions

In addition to some bolts having large unbonded lengths, further examination of the data showed that there appeared to be differences in the bolt characteristics in different parts of the mine (Figure 4). This plot shows the distribution of unbonded lengths in three sections along a roadway. The mean unbonded length in each section is circled, and the length of the horizontal line indicates the spread of the data in the region. Although the spread of all three sets of data is similar, it is clear that two of the regions have similar mean unbonded lengths of around 250-300mm while the third has substantially shorter unbonded length which is much nearer the zero unbonded length of the original installation specification.

Exception testing

GRANIT tests were also taken on a number of bolts to determine load. Once tested it is possible to identify bolts significantly different from others in the vicinity i.e. exception testing. It is then possible to deduce cause, such as local geology, adjacent workings or installation practice and procedures, as a basis for remedial action.

In this context, exception testing at Thoresby resulted in the identification of a number of bolts with inordinately long unbonded lengths/low loads and bolts diagnosed as having very high loads. A picture of the loading in the roadway can be built by plotting response frequency/load against location of the bolt(s) in the roadway. Figures 5 and 6 show the raw frequency data for over 250 bolts, taken in the two main access roadway areas and panel roadways respectively. Although there are large clusters of data between 800Hz and 900Hz in two sections of the plots, there are clear exceptions below 600Hz and above 1000Hz. These correspond, respectively, to bolts with very long unbonded lengths/low loads and very highly loaded bolts.

Figures 5 and 6 also show clear trends of increasing response frequency along the roadway. These corresponded precisely with increasing visible loading in the roadways and associated increasing deformation along the roadways in those locations.

In another aspect of these trials, two different groups of bolts of different age were tested to determine the effect of bolt age on vibration response, if any. The bolts tested in the main access roadway were approximately 30 months old and those in the panel roadway approximately two months old. The vibration response signatures were seen to be internally consistent in both age groups and there was no discernible difference in signature between the two groups, except where unbonded length is concerned, as noted above.

Discussion and implications

By monitoring bolts, GRANIT data can be used to assess changes in their loading pattern, which would give indications of incipient roof movement. Regions of higher indicated load can then be checked on a more regular basis, offering potential for improved safety.

One early indication from the testing of 264 bolts in the two different areas of Thoresby Colliery was that although the bolts were seen to be in compliance with installation quality control criteria and with regulations, GRANIT proved there was a high proportion of bolts with a measurable free length, notwithstanding full encapsulation to the mouth of the hole.

These bolts were not intended to be passive bolts but in practice were acting as such. Also, perhaps not coincidentally, these bolts were clustered in an area of roadway showing marked deformation and heavy loading. In civil engineering practice, active bolts would be specified for typical roof support bolting/anchoring system, with a significant degree of pre-stress applied.

It should be noted that the benefits of pre-stress can still be obtained, even though full encapsulation to allow transmission of the horizontal stress field at all horizons in the bolted zone is required. The use of GRANIT has shown that this is an issue the industry needs to address, as the current non pre-stress anchor methodologies are not gaining the full degree of possible support mobilisation, which is obtainable by the application of pre-stress.

Future developments

GRANIT is being continuously developed. Initial tests on single tendon strand anchorages similar to the cable bolts used in mining have proven successful(6). Testing has also taken place at a number of sites to assess the application of GRANIT to large bar bolt anchorages and the initial results show the system is applicable.

The system has been streamlined to make it easily portable/operable by one man. The impact device has been reduced in size and weight to improve handling, ease of connection to the bolt head, and to improve productivity. This will allow the system to be run readily from small compressed air bottles. In parallel, the data acquisition system is being miniaturised by replacing the laptop computer currently used, with a handheld device with sufficient memory to allow data from several shifts to be stored before downloading.

Research is also being undertaken to extend the capabilities of the system to assess total bolt length, the presence of broken or bent bolts and loss of tendon cross section. One very important aspect of this is the fracture failure of the anchor tendon and this can be due to a variety of reasons. Loss of section due to corrosion, failure due to overloading and stress corrosion cracking are possible causes. Research work and practical experimentation is proceeding into determination of tendon length and early results are very promising. Work is also proceeding into the application of GRANIT to the detection of loss of section due to corrosion, and to improve the scope and accuracy of the mathematical model. Research funding and potential trial sites are being sought to extend the use of GRANIT to multi-strand anchorages.

Conclusions

The results presented show the GRANIT system can identify rockbolts with very high loads and/or poor bonding in an underground environment. The assessment of load under the head of the bolts provides an indication of incipient vertical roof movement.

Long term monitoring of the load would permit trends in the load to be observed and potential problems to be anticipated. Exception testing to identify anchors without the normal envelope of performance, loading or specification is proven.

It complements and enhances the monitoring spectrum between visual examinations and lift off tests and/or special instrumented anchorages.

Related Files
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6