Sunteți pe pagina 1din 10

Green Paper 101

April 2013

Page 1 of 10
Real Aperture Radar for Safety-Critical Slope Monitoring
Patrick Bellett, PhD and David Noon, PhD
GroundProbe Pty Ltd
Brisbane, Australia
Abstract
This paper explains the radar physics evidence which demonstrates why Real Aperture Radar (RAR)
is the best radar technology to provide robust alarming for monitoring slope behaviour. This is the
fundamental premise upon which the development of the safety-critical monitoring products provided
by GroundProbe is based.
Introduction
As the inventor of slope stability radar, GroundProbe has specialised in continuously redefining the
most suitable radar architecture to perform effective slope risk management for safety-critical mining
activities. The safety-critical aspect requires careful technology choices to ensure that trades in
system performance are made in areas that have the least impact on performance. The result is a
system that is able to provide robust data that correctly locates areas of slope instability and reliably
assesses and produces user-defined movement alarms.
A decade of working with the global mining industry in some of the most remote, challenging and
varied environments, has validated the real aperture antenna design approach for safety-critical
monitoring. It is a keep it simple solution which has proven to be effective.
This paper explains the science of the simplest implementation of a real aperture radar using a
parabolic dish antenna. This type of real aperture radar antenna is designed to do all the hard work,
providing the best capability for robust alarming, simple point-and-click functionality and
straightforward processing for safety-critical slope monitoring.
Safety-critical applications and requirements
For open-cut mining, the most demanding application for a slope stability monitoring system is safety-
critical monitoring, where the safety of mining personnel and equipment could be immediately at risk
[1]. Compared to non-safety-critical applications, by nature safety-critical monitoring imposes more
demand on technology and system requirements to ensure the most correct and robust slope failure
measurement and alarming is provided. This is best achieved with a radar system that can monitor the
slope with the highest degree of pixel independence, using negligible processing complexity and data
Green Paper 101
April 2013

Page 2 of 10
load, and imposing minimal constraints on the positioning of the monitoring system within the pit,
relative to high risk failure areas.
In contrast, very shallow pit slopes, slopes that are not moving or are moving very slowly with
expected behaviour, or slopes that do not pose operational risk to people, equipment or mine
infrastructure, may not require real-time, safety-critical monitoring. However, for overall pit risk
management, these types of slopes may need to be monitored in a background sense (i.e., for
campaign monitoring or long-term trending analysis). Background monitoring is conducive to a
number of fixed systems that monitor broad areas over long timeframes, for example, total station
prism monitoring systems [2] [3]. A summary of the application spectrum for slope stability monitoring
is provided in Table 1 to explain the different monitoring paradigms.
GroundProbes focus is on safety-critical monitoring, so all the GroundProbe products, including WAM,
SSR-T and SSR-XT, and services are based on real aperture radar technology and its data. For non-
safety-critical (background) monitoring, complimentary technologies exist, e.g., total station prism
systems and Synthetic Aperture Radar (SAR) [4] [5] [6].
Table 1: Application spectrum for slope stability monitoring
Safety-critical

Non-safety-critical
Requires real-time monitoring
Does not require real-time monitoring
(background monitoring)
Highest immediate safety consideration Lowest immediate safety consideration
Operational critical decisions & alarming No alarms to operations
Real-time instant slope monitoring Long-term slope trend monitoring analysis
Rapid failures Slow, large failures
High mobility flexibility of locations Semi-permanent to fixed locations
Shorter monitoring timeframes Longer monitoring timeframes
Shorter monitoring distances Longer monitoring distances
Focussed area coverage Broader area coverage
Alarming performance benchmark for safety-critical monitoring
No monitoring system is perfect; hence, all monitoring systems have the potential to miss failure
alarms and produce unwanted failure alarms. This philosophical assessment is summarised by the
arbitrary system alarm performance table in Figure 1a, where the potential alarm confusion is
adjustable (as noted by the two-headed arrows) based on the system design and associated trades.
This type of binary classifier table or confusing matrix is used in many fields; for instance, they are
used in medical devices and data mining applications and are often complicated by the use of
confusing positive-negative logic terminology (e.g., false-positive, true-positive, false-negative etc.) [7].
Green Paper 101
April 2013

Page 3 of 10
Our approach simplifies the alarm matrix interpretation by making the terminology more applicable to
the mining industry. The most important term is the missed alarm, which is comparatively intuitive,
meaning the system did not produce an alarm resulting from actual wall behaviour (i.e., a false-
negative). Missed alarms have a serious safety cost with the potential for loss of life or injury to
people. Missed alarms also have a financial cost with the potential for damage to equipment and mine
infrastructure. The alternative is an unwanted alarm, which is an alarm produced by the system when
it should not ideally have produced an alarm (i.e., a false-positive). However, unwanted alarms are a
lesser evil, associated with an inconvenience, which usually has a financial cost but no safety cost.
For example, mining is stopped unnecessarily for the duration taken to investigate the alarm event
before operations can continue.
Unwanted alarms are also caused by equipment failure and user error (e.g. alarm thresholds not set
or inappropriately set), which are independent to choice of technology. We prefer to use the term
false alarmsfor these technology-independent unwanted alarms, which require separate policies,
procedures, training, and support to be managed effectively.
For a safety-critical monitoring system, it is paramount to design the system to minimise the chance of
missed alarms. It is equally important to understand the impact of unwanted alarms on the overall
system performance. For example, automatic and systematic suppression of unwanted alarms is
highly undesirable in a safety-critical system as this inevitably comes at the high cost of increasing the
missed alarm potential. For slope stability radar, a pertinent example of unwanted alarms relates to
atmospheric variations in the data. It must be emphasised that a careful balance of system design
trades is required to minimise missed alarms.


(a) (b)
Figure 1: Alarming performance characterisation for slope monitoring radar. (a) An arbitrary system
characterisation demonstrating how the potential for missed alarms and unwanted alarms can vary with
technology choice; (b) the benchmark safety-critical system characterisation that absolutely aims to minimise
missed alarms.
Green Paper 101
April 2013

Page 4 of 10
In general, it is well understood from theory and practice that efforts to lower missed alarms will result
in higher unwanted alarms, and vice versa. The classic theoretical analysis was developed by Radar
engineers during World War 2 in trying to discriminate true targets from false targets on radar operator
screens [8]. For slope stability monitoring, a practical example is that a low deformation threshold
results in less missed alarms at the expense of more unwanted alarms and, conversely, a high
deformation threshold results in less unwanted alarms at the expense of more missed alarms.
In the petroleum industry, a major turning point in the understanding and acceptance of safety-critical
alarming practices occurred in 1988 following the Piper Alpha North Sea oil rig disaster [9]. This
example is especially noteworthy as 167 people died as a result of the gas leak fire. It seems that
before the disaster, it was common practice to suppress unwanted alarms to maximise production.
However, after the disaster, unwanted alarms were considered acceptable practice and stopping
production for an unwanted alarm was considered good practice. Alarms need to be managed
operationally to ensure no alarm is missed. Consequently, false alarms become unacceptable
because the cost of a false alarm increases. As discussed above, this highlights the importance of well
trained personnel, good procedures and maintenance plans to help manage the overall risk profile.
The safety-critical benchmark for slope stability monitoring aims to absolutely minimise the potential
for missed alarms and, further not to automatically suppress unwanted alarms at the expense of
increasing the potential for missed alarms. This is illustrated by Figure 1(b).
Real Aperture Radar
Slope stability monitoring radars use a phase based technique called interferometry to measure small
changes in range (i.e., in the line-of-sight direction towards the radar) associated with pre-cursor
movements to mine slope failures. These small changes in phase (corresponding to the appropriate
fraction of the radar wavelength in millimetres) are accumulated over time for each subsequent scan
and converted to millimetres of displacement.
Before useful deformation measurements can be made, especially for safety-critical applications,
these radars require a good spatial filtering method for image formation that can reliably map or
characterise the physical 3-D shape of the slope in a way that is simple and intuitive for easy
interpretation and alarm evaluation. The simplest way of imaging a complex 3-D surface is to use a 3-
D radar. The most robust 3-D radar implementation uses a two-axis, mechanically scanned radar dish
antenna to raster scan a real pencil-beam over the wall area, coupled with using range information
from the radar. This method is illustrated in Figure 2.
Green Paper 101
April 2013

Page 5 of 10

Figure 2: A real aperture radar with a mechanically steered dish antenna forms a very effective 3-D spatial filter,
shown by the radar beam footprint on the wall surface, for robust imaging and deformation mapping. (Adapted
from [6])
The parabolic dish antenna is the most simple and trusted method of consistently forming a highly
robust, reliable and instantaneous real pencil-beam with a narrow conical beam shape. The angular
beamwidth of the pencil-beam is related to the size and shape of the dish antenna. To achieve a
narrower pencil-beam, a larger dish diameter is required [10].
The range information is provided by the radar signal bandwidth and is related to the range resolution
of the system in metres [10]. The range resolution of the radar combines with the real pencil-beam,
afforded by the parabolic dish antenna, to provide the most effective 3-D spatial filter for imaging a 3-D
wall surface. This means the RAR is able to generate its own 3-D digital terrain map (DTM) that
characterises the scanned pit geometry, and uses this to robustly segment the wall surface into
reliable, independent pixels for evaluation and alarming.
Figure 3(a) and Figure 3(b) show how the raster scanning process of a real aperture radar forms a
regular angular grid of pixels that represents the 3-D wall surface as a 2-D image. In particular, Figure
3(b) shows how this simple image registers well with a photographic image of the wall, which
inherently provides a uniquely intuitive visualisation format for complex radar data. Figure 3(b) also
shows how each image pixel represents a definitive area on the 3-D surface of the wall with 3-D
coordinates(Az

, El

, Rongc).
It is important for safety-critical applications to demand a very high degree of pixel area selectivity or
pixel independence to ensure deformation measurements are not contaminated and are correctly
located or mapped against the actual 3-D wall surface. Figure 3(c) shows how pixel deformation errors
caused by interference anywhere outside the immediate pixel angles are all but eliminated by the RAR
spatial beam-filter.
Green Paper 101
April 2013

Page 6 of 10
(a) (b) (c)
Figure 3: RAR raster scanned image formation from independently sampled areas of the wall. (a) shows how an
arbitrarily shaped rough wall surface is sampled in angle, (b) shows how 3-D surface reflection voxels are
represented as 2-D image pixels, and (c) shows how the RAR pencil-beam achieves a very effective spatial filter
for each image pixel to achieve independent measurements.
Figure 4 illustrates how all the independent pixel phase measurements are compared at each scan
and accumulated to show where the wall is deforming. The amount of deformation is represented
visually by a colour palette gradient.

Figure 4: RAR produces independent scan area pixels that allow interferometric phase accumulation to clearly
indicate areas of the wall that are deforming. The deformation mapping is intuitively registered with the
photographic image.
The effectiveness of the RAR spatial beam-filter simply relates to the inherent sidelobe suppression
capability of a real aperture antenna; where the radar beam pattern is derived from the free-space
multiplication of the two real antenna beam patterns, one on transmit and one on receive [10]. This
capability is unique to the RAR approach and is fundamental in providing safety-critical performance
by not missing alarms. For example, very bright reflection from machinery operating within the vicinity
of the scan area will be effectively filtered and not swamp or contaminate any other pixels in the scan.
The RAR pixel filter is illustrated further in Figure 5(c), where it is shown to have soft edges as a result
of the radar pencil-beam weighting; that is, the radar sees the wall better at the centre.
Green Paper 101
April 2013

Page 7 of 10
All antenna systems produce some degree of unwanted sidelobe structures outside the main-beam
pattern as a result of aperture edge effects and errors (e.g., as would result from damage to the dish
causing the shape of dish surface to change). For safety-critical monitoring of mine wall surfaces, the
distribution of antenna sidelobe energy needs to be well controlled, otherwise it will unknowingly
impair the monitoring system and alarm performance. Excellent sidelobe performance is easily
achieved with a real aperture parabolic dish antenna, which provides a very good spatial filter as
shown in Figure 5(c) [10] [11].

(a)

(b) (c)
Figure 5: (a) shows the wall scan area with the same identified pixel seen in Figure 3(b); (b) shows a zoomed
view of this pixel; and (c) shows the highly effective RAR spatial filter with soft edges for the same pixel.
The 3-D RAR system provides the convenience of pit geometry independence that relaxes constraints
on where the radar needs to be positioned. Figure 6 illustrates this point with a series of simple wall
geometries of varying overall slope angle. The RAR can still monitor the wall in 3-D irrespective of the
pit slope angle or shape from most practical locations.
However, the limitations of the 3-D RAR system are increasingly exposed as the operating range of
the system is increased. This is because the largest practical sized dish antenna limits the effective
operating range of real aperture radars to several kilometres, beyond which bench scale failures can
no longer be resolved. However, it should be recognised that multi-bench scale failures can still be
resolved beyond several kilometres.
Green Paper 101
April 2013

Page 8 of 10

(a)

(b) (c)
Figure 6: The 3-D RAR system can image the wall irrespective of the pit slope angle or shape. Three simple
examples are used to illustrate geometry independence: (a) a very steep wall; (b) a reclined wall; and (c) a
shallow angle wall.
Conclusions
We conclude that the principles of radar physics applied to safety-critical slope monitoring determines
Real Aperture Radar (RAR) as being the most robust radar technology for high-confidence, real-time
alarming. This is because RAR antenna does all the hard work in forming a 3-D spatial beam with the
lowest sidelobes to directly measure slope movements at each independent pixel position. Further, the
3-D RAR provides mine pit geometry independence; this, in turn allows a simple point-and-click setup
functionality with straight-forward processing for robust alarming.

Green Paper 101
April 2013

Page 9 of 10
References
[1] N. Harries, The use of Slope Stability Radar (SSR) in managing slope instability hazards,
Geomechanics, pp. 53-54, J anuary/Februrary 2008.
[2] M. J . Little, Slope monitoring strategy at PPRust open pit operation, in The South African
Institute of Mining and Metallurgy, International Symposium on Stability of Rock Slopes in Open
Pit Mining and Civil Engineering, Capetown, South Africa, 2006.
[3] A. Vaziri, L. Moore and H. Ali, Monitoring systems for warning impending failures in slopes and
open pit mines, Nat Hazards, vol. 55, pp. 501-512, 2010.
[4] P. Mazzanti, Remote monitoring of deformation using Terrestrial SAR Interferometry (TInSAR,
GBInSAR), Geotechnical News (www.geotechnicalnews.com), p. 28, March 2012.
[5] F. Bozzano, P. Mazzanti and A. Prestininzi, A radar platform for continuous monitoring of a
landslide interacting with a under-construction infrastructure, Italian Journal of Engineering
Geology and Environment, vol. 2, pp. 35-50, 2008.
[6] E. L. McHugh, E. G. Long and C. Sabine, Applications of ground-based radar to mine slope
monitoring, in ASPRS Annual Conference Proceedings, Denver, Colorado, 2004.
[7] I. H. Witten and E. Frank, Data mining: practical learning tools and techniques, Second Edition
ed., San Francisco, CA: Elsevier, 2005.
[8] J . Fan, S. Upadhye and A. Worster, Understanding reciever operating characteristic (ROC)
curves, Canadian Journal of Emergency Medicine, vol. 8, no. 1, pp. 19-20, 2006.
[9] Piper Alpha, Wikipedia, (n.d.). [Online]. Available: http://en.wikipedia.org/wiki/Piper_Alpha.
[10] M. I. Skolnik, Radar Handbook, 3rd ed., McGraw-Hill, 2008.
[11] I. D. Longstaff, Comparing real beam and synthetic aperture techniques for Slope Stability
Radar, Whitepaper, University of Queensland, Australia, 2011.

Brief Biographies of the Authors
Patrick Bellett is the Principal Research Engineer at GroundProbe. Pat is responsible for research
activities, including early proof-of-concept developments for safety monitoring radar technologies. Pat
has a PhD in Electrical Engineering in the field of radar antennas and has been involved in advanced
radar R&D for mining applications since 1995. He has also worked overseas and has experience
in distributed process control and alarming systems in the Canadian oil and gas industry. He holds a
Bachelor Degree in Electrical Engineering and a Research Master Degree in Engineering Science
from the University of Queensland, Australia. Pat is a member of the US Institute of Electrical and
Electronic Engineers (MIEEE).
David Noon is an inventor of the patented Slope Stability Radar technology, a founder of GroundProbe
and leads its R&D and commercialisation activities. David has a PhD Degree in Electrical Engineering
Green Paper 101
April 2013

Page 10 of 10
in the field of high-resolution radar, and a first-class honours Bachelor Degree in Electrical
Engineering from the University of Queensland, Australia. He has pioneered the R&D and
commercialisation of advanced radar technologies in mining applications since 1993. David is a Fellow
of the Australian Academy of Technological Sciences and Engineering (FTSE), a Fellow of the
Australasian Institute of Mining and Metallurgy (FAusIMM), a Senior Member of the US Institute of
Electrical and Electronic Engineers (SMIEEE), and a Registered Professional Engineer in Queensland
(RPEQ).

S-ar putea să vă placă și