ADAM Technology

Four different monitoring methods were used to provide parameters that were easily measurable in the field as well as in numerical models:

• Closure measurement (Tape Extensometer) – tried and tested method of sidewall deformation measurement intended as a backup check of other deformation measurement techniques (photogrammetry and laser). Results are directly comparable with numerical modelling deformation predictions.

• Closure measurement (photogrammetry) – modern technique for creating a 3D georeferenced model of the excavation skin, using a mosaic of overlapping high-resolution photographs. Successive scans can be compared with each other to gain a 3D profile of the excavation deformation, for comparison with numerical model predictions.

• Shotcrete liner strain – vibrating wire strain gauges embedded in the liner, directly measuring the circumferential strain within the liner shell.

• Cablebolt load – SMART cables (Hyett et al, 1997) installed in a fan around the excavation perimeter, to directly measure strains in the cable element and indirectly indicate the extent of the rockmass yield zone.

3DM Analyst Mine Mapping Suite and its associated Underground Field Kit (Figure 1) were selected as the photogrammetric system to use for measuring tunnel skin deformation.

Figure 1. Photogrammetry field setup, consisting of tripod mounted digital camera and floodlighting.

All measurement systems were able to detect convergence, although 3DM Analyst detected more convergence than the others and was found to be sensitive to the control point specification, as would be expected if control points were the sole method being used to georeference the surfaces at different epochs. (ADAM recommends tying projects together using relative-only points when higher relative accuracy is required, as is often the case with monitoring projects.) The paper found that further work was required on the control point survey methodology.

Figure 2. Superimposed photogrammetry surface models indicating areas of convergence.

Figure 3. Cross-section through the surface models showing relative deformation.

The paper concludes:

"The monitoring trial has been successful in delivering a dataset of sufficient quality to be used in calibrating a numerical model. Key performance elements of the system have been captured, namely tunnel deformation, cable support load and liner strain. The data obtained suggests that the system is behaving (conceptually) as would be anticipated."

The paper and associated presentation explore several uses of the 3D data generated by 3DM Analyst Mine Mapping Suite, including using it to update the structural models:

The surfaces created for the structural model use information from the 3D images which is millimetre accurate, as well as drill hole data and seismic data to constrain their location. As such, if a cut intersects a major structure, then the data is added to the model as soon as possible, thus resulting in the models always being up to date. The surfaces are created using true point data, with all varying data types in separate layers so that the data integrity of the surface can be assessed, a bounding polygon constrains the surface. By using only data points to constrain the location of the surface (Figure 3), irregularities common with using polygonal modelling are negated. This setup enables surfaces to be rerun with new data within five minutes by any member of the geology team. Any significant variations to the models are then communicated to the engineers to ensure that all development is adequately supported and future development is located in the best position possible. Every time a surface is updated, all of the 3D images that intersect the surface are loaded to enable the style of the structure to be assessed across its extent. This is particularly important considering that many faults preserve varying levels of deformation depending upon the rock types involved.

Convergence monitoring is also mentioned:

Figure 1. Convergence monitoring.

The paper concludes:

"The successful mining of underground deposits requires detailed geological interpretations based upon sound geological data. In mines that use shotcrete and bolts for their primary ground control regime such as Cosmos, the ability to assess raw data such as the AT 3D images provides countless benefits. There is simply no comparison between using the 3D images, as opposed to poor quality 2D face photos. The use of 3D images enables geologists to assess the raw data at any point, without the imposition of time constraints. Thus, entire drives can be assessed by simply turning on the computer, no travel time required, with the added benefit of being able to incorporate all other mining data into the interpretation process."

A single strip of nine images were captured 50 m from the rock face with a 50 mm lens:

Figure 1. Project plan.

Figure 2. Photographing the rock face.

Figure 3. Mapped structures.

The paper concludes:

Digital photogrammetry worked quite well on this project. The discontinuity measurements from the digital photogrammetry model were used in the stereonet analyses of the kinematic potential for plane, wedge, and toppling failures. The measurements were also used in optimizing the azimuth of horizontal drains proposed to reduce water pressure within the rock mass and improve cut slope stability. Having the coordinates of each measurement permitted analysis of the variability of bedding dip angle and dip direction along the curving highway alignment. The digital model was also useful for locating the possible outcrops of clay seams encountered in test borings.

This project illustrated some of the advantages of using digital photogrammetry for rock cut slope design. Specific advantages include the following:

1. Data can be collected safely from challenging to reach areas with no rappelling or free climbing.

2. Exposure to traffic, rockfalls, and inclement weather is significantly reduced.

3. A robust set of measurements can be collected in a short period of time.

4. There is no need to set up and measure traverses since the software provides coordinates of each measurement.

5. Discontinuities are measured across a greater area of their extent than can be measured by hand.

6. Discontinuities with limited exposure are easily measured by digitizing their trace across the outcrop model.

7. Measurements are fitted to the digital image of the rock face, which can be rotated and viewed edge-on to assess the closeness of the fit.

8. Having the coordinates of each discontinuity measurement facilitates analysis of the variability of bedding dip and dip direction along the roadway alignment.

9. Human error in recording the measurements is reduced since the measurements are recorded by the computer.

10. Shallow dipping (<5 degrees) joints, which can be very challenging to measure with a compass, can be easily measured by digitizing their trace along the digital model.

Although useful for quickly and safely collecting discontinuity orientation data required for kinematic analyses, digital photogrammetry may not permit measurement of fallen block sizes and observation of seepage conditions, joint roughness, weathering, slickensided surfaces, and joint in-fillings needed for other analyses factored into the design of a rock cut.

At the Golden Rocks conference in 2006 a workshop was conducted to investigate the state-of-the-art in laser scanning and photogrammetric methods for rock face characterisation.

Vendors were required to demonstrate how a 3D model of a rock face could be captured using their systems during the workshop, and were required to submit an analysis of the rock face to the organisers for review in November, 2006.

ADAM Technology was the only vendor who created the 3D model of the entire rock face in real-world co-ordinates during the workshop — in fact, we did it twice, using two different lenses and two different techniques to show they delivered the same result. It took just over 10 minutes using one technique and just over 20 minutes using the other.

ADAM Technology was also the only vendor who submitted the analysis by the original deadline. The deadline was extended by three months and even then only one other vendor submitted any data at all.

The organisers evaluated the submitted data and found that when compared with 12 control points not used for orientations, ADAM's terrain model was three times more accurate on average (with the largest error being 1/5^th as large) and four times more dense, but required just 1/9^th the time to generate (90 minutes vs less than 11 minutes).

Figure 1. 3DM Analyst's terrain model.

Figure 2. Other vendor's terrain model.

Thanks to the "averaging" nature of discontinuity orientation measurements, however, where many DTM points are used and the best fit plane found, the actual joint orientations, set means, and trace lengths agreed extremely well. The organisers concluded that "the obtained results indicate that digital photogrammetry yields reliable and reproducible results when applied to rock mass characterization. Digital photogrammetry is thus a mature enough technology that can be used with confidence in the profession provided care is taken to follow the guidelines provided by the Presenters in their papers and by the authors in the introductory paper to this report."

In early 2007 Rio Tinto Iron Ore conducted a trial to compare 3DM Analyst Mine Mapping Suite with Sirovision, which they had been using for geotechnical engineering for several years.

Despite the extensive experience they already had using Sirovision, and just a couple of hours of training on 3DM Analyst, they found that:

"[T]he image capture using ADAM technology is approximately 1/4 of the time taken with SIRO. Data processing (for projects) as much as 1/5–1/6 of the time as compared with existing SIRO techniques."

They also reported:

"It should also be mentioned that the ADAM user interactivity was rapidly picked up, with simple use of available tutorial information and their relatively automated systems. In comparison, the existing in-house SIRO training and competency has been recognised as not only thorough but also arduous, limiting the potential number of staff proficient with it at any one time."

Please contact ADAM Technology for a copy of this report.

In 2006 Zostrich Geotechnical published a presentation comparing digital photogrammetry using 3DM Analyst to standard geologic mapping techniques.

They found that a job requiring a total of 36 staff-hours could be completed using 3DM Analyst in just 3.5 hours. They also observed that by using the 3D view of the DTM in the software, "The engineer can safely examine the potential failure area in great detail from a multitude of angles. It is possible to literally 'fly' around and above potential rock failure blocks, even those in active failure. [...] While physical examination of the block may be required at some point, this method allows for detailed examination and analysis of the situation without the rock scaling and rope work normally required to assess the situation."

They concluded:

"Digital photogrammetry, and associated rock analysis with the assocaited DTM's and images increase the efficiency, and accuracy of any rock design where large or poorly accessible areas must be mappped, great geologic detail is required, or where dangerous mapping situations exist."

In early 2005 Roche Mining conducted a trial to evaluate the suitability of 3DM Analyst Mine Mapping Suite for highwall coal seam mapping, dump profiling, and blast profile mapping.

They compared the results using a range of different cameras, from the Canon EOS 10D digital SLR down to $300 compact digital cameras, as well as a laser scanner being operated by an independent contractor.

They found that the 3DM Analyst data was within 0.3% of the values obtained using conventional surveying techniques, while the laser scanner was only within 1.2%. The author reported in private correspondence that the fieldwork for the laser scanning was 3–4 hours with two people plus 30 minutes of processing, while the fieldwork for 3DM Analyst was 2 hours for one person (capturing photographs using multiple techniques and using many more control points than usual, for testing) plus 1.5 hours of processing.

Reprocessing those same images two years later, with the latest version of the software on a current PC, takes only 20 minutes — a five-fold improvement in productivity! This is one of the big advantages of a software-based system that relies on cheap and readily-upgraded hardware over a technology that relies on expensive instruments, like laser scanning does — every year the system gets better, rather than increasingly obsolete.

In 2004 BHP Billiton-Mitsubishi Alliance (BMA) conducted a trial to compare 3DM Analyst with Sirovision for the purpose of surveying for resource model/reconciliation, highwall mapping (lithology, geology, selected horizons), and resource modelling (top/bottom of seams, horizons, faults, dykes).

Each vendor was invited onsite for a day to perform the field work and was later required to submit their data for comparison. Both teams used identical cameras (Nikon D1x's, both originally supplied by the CSIRO) and identical lenses for the main portion of the trial, allowing direct comparisons between the two packages to be made. The result is an incredibly extensive report on the outcome of this trial.

Using identical cameras and lenses the 3DM Analyst data was found to be over twice as accurate on average, with the largest error being less than 1/3^rd the largest error of Sirovision. The vertical accuracy was 3.4 times better.

The report also states:

"BHPB Technology has generated a large number of 3D images for BHPB Iron Ore operations during the last two to three years. Some of the recent images have been rerun with 3DM Analyst. The range of camera to image distances was 450 m to 850 m. It was found that in every case the 3DM Analyst images were significantly easier and faster to be generated and the resulting 3D images were more accurate than the Siro 3D images."

The report concludes:

"The results of the field trial at Goonyella have indicated that the Adam Technology software is more accurate, faster, more flexible, easier to use, more robust and more rapidly developing, provides better quality models, has better software support and requires less training. Based on technical considerations of the trial results, Adam Technology's 3DM Analyst plus CalibCam softwares provide the best solutions for BMA's highwall mapping."

The report also makes a number of recommendations for further enhancements, noting that "During the test period a number of recommendations were made to make the 3DM Analyst software more suitable for large mapping projects outlined in the scope of work. Adam Technology was very proactive and started working on the recommendations immediately. As a result, the software has improved significantly. Some of the remaining enhancement suggestions are listed here." ADAM is very proud of the fact that every one of the recommended enhancements were added to the software within two months of the report's release!

The download includes the original report, a PowerPoint presentation containing the figures referred to in the report, and ADAM's own report on the trial, referred to in the BMA report as "Appendix 4".