Slope Stability 2020 - PM Dight (ed.

)
© Australian Centre for Geomechanics, Perth, ISBN 978-0-9876389-7-7
doi:10.36487/ACG_repo/2025_23

BHP Western Australia Iron Ore geotechnical open cut slope

design system: a simple pragmatic process for slope
risk decisions

A Haile BHP, Australia

D Ross BHP, Australia
A Maldonado BHP, Australia
M Neyaz BHP, Australia
C Rajbhandari BHP, Australia

Abstract
A transparent, pragmatic geotechnical design system is outlined that presents a selection of risk options with
associated risk/reward for decision makers. Risk options termed ‘robust’, ‘balanced’ and ‘aggressive’ have
been defined appropriate to ‘critical infrastructure’, ‘typical industry’ and ‘low risk’ mining environments
(where the safety risk and the consequences of failure on the budgeted mine plan are acceptably low),
respectively. The geotechnical model includes ‘most realistic’ and ‘reasonable lower case’ conditions.
A ‘realistic’ design principle requires reporting a Factor of Safety on the realistic case, rather than to reduce
design inputs due to uncertainty. Uncertainty is transparently covered by the ‘lower case’ in sensitivity
analyses. Indicative probabilities of failure are estimated and a simple empirical tool estimates the
consequence of failure in terms of the area of mining floor impacted. These together with indicative value or
tonnage estimates are presented to the decision maker (risk owner), with a selection recommendation. Post
decisions, designs including any residual hazards are passed to the operational engineers to design risk-based
slope monitoring, to ensure operational safety, and reconciliation programs, as required. Adoption of the
realistic principle has facilitated risk owners taking decisions based on a more transparent presentation of
risk since 2014 and made a material contribution to a step change slope angle increase in the mines of BHP
Western Australian Iron Ore.
Keywords: design, uncertainty, failure, risk
Confidential. © BHP 2020. No information in, or part of, this document may be reproduced or disclosed
without the express written permission of BHP.

1 Introduction
BHP Western Australia Iron Ore (WAIO) own and operate open cut iron ore mines in the Pilbara region of
Western Australia. The pits are predominantly in Archean folded banded iron formations with some tertiary
age sediments. Approximately 40 active cutbacks are mined at any one time, with ultimate design depths
between 50 to 450 m but typically 100 to 150 m. The average mining depth across the portfolio in 2018 was
only approximately 50–75 m. From 2000 to 2014, geotechnical design was outsourced to various parties,
with differing design principles and variable levels of transparency on strength evaluation. Since 2014,
designs were carried out to a common design philosophy with evolving design principles. This was largely
based on common industry practices, however, it mandated realistic strength estimates for the most credible
case and sensitivity analyses on lower cases with corresponding lower case acceptance criteria. In this paper,
this slope design process for approving pits for construction is outlined and the high level results of
implementation to date. Thereafter there is a discussion and conclusion.

Slope Stability 2020 415

BHP Western Australia Iron Ore geotechnical open cut slope design A Haile et al.
system: a simple pragmatic process for slope risk decisions

2 Process overview
Figure 1 outlines the geotechnical design process, stepping through from data collection to model
development, analysis, risk assessment and slope option selection. The main points of which are drawn out
in the subsequent sections.
The focus in this paper is predominantly on approving designs for construction, particularly:
• A geotechnical model that provides realistic and credible lower case estimates for key parameters.
• Analysis that includes indicative probability of failure (PoF) estimates.
• Slope risk options with varying acceptance criteria.
• A simple assessment of risk and relative value of the options.
• Accountability for slope decisions.

3 Data collection
Each discipline (geotechnical, geology, hydrology) is accountable to collect data as an input to models that
are fit-for-purpose for the design stage. Investigation designs are shared for cross-discipline comment to
optimise data collection. For the rock mass model, for pits for construction, this typically involves a
site-specific geotechnical ground investigation that aims to generate suitable data on the geotechnical units
that are material to the design and to cover a reasonable geographic range. This typically includes site-specific
diamond core and televiewer of reverse circulation (RC) and diamond holes.

4 Geotechnical model
As far as reasonably practical, the geotechnical model is based on the concept of a realistic design principle.
The realistic design principle aims to present the most realistic representation of the in situ ground conditions
that will control stability. This is assessed irrespective of data quantity or quality. Any uncertainty in this
assessment is then covered by an estimate of what a lower reasonably credible case or ‘lower case’ estimate
could be. The lower case is not the absolute minimum condition of a parameter or a combination of multiple
lower case parameters, unless these are correlated. Notwithstanding that the combination of these multiple
lower case conditions may occur, their probability of occurrence is considered to be sufficiently low as to not
influence the design decision.
The geotechnical model is the synthesis of the major stratigraphic/structural, hydrological, rock mass and
minor structural models.
The geoscience team provide major stratigraphic and structural (fault) models, typically based off 50 × 50 m
exploration RC drilling. Confidence solids for the model are provided. Confidence is based on a rating matrix
including drillhole spacing, interpretation confidence (logging, geochemistry, geophysics), supporting
televiewer data and major structure intercepts.
Hydrology provides estimates of the groundwater pressure in 2D line sections or occasionally 3D models and
representation again of the most realistic and lower case estimates.

416 Slope Stability 2020

 Planning Collect Geotechnical Stability Slope Slope
Cycle Data Model Analyses Design Decision

Geology Pit
Model Design Design
• Risk Slope Risk
E

Slope Stability 2020

Options Profile

Design Ground • Uncertainty

Investigation • Hazards
Geology • Additional Options
Model Benefit / F
Stability
Measures Consequence
Fieldwork
5Y Mine Proposal Major
Plan B
Structures Risk Decision G
2Y Mine
Schedule Design Report
Exploration,
Geomet, Hydro Hydro
A Review Model C
Pit Design

Prepare Review
Finalise Minor Geotechnical D
Geotech Budget Proposal Model
Structure
and Work
Schedule Design Check
Fit for
Purpose Pit Review
Log Rock
Geotech 2Y Strength Design Response
drillholes, Mass
Budget and Model Review
mapping Hazard
Schedule H
Acceptance

QAQC Solids and

Review IRA OSA Risk
Slope
Response Analyses Management
Angles
Implementation I
Lab Testing Plan
Rock
Mass Batter Analyses
Approved Model Geotech Master
Logs Area Upload

Project
Closure

Figure 1 Western Australia Iron Ore slope design process overview

417
Safety and risk management
BHP Western Australia Iron Ore geotechnical open cut slope design A Haile et al.
system: a simple pragmatic process for slope risk decisions

The geotechnical discipline constructs the rock mass and small-scale structural model. Where sufficient data
exists, selection of central and lower cases, is typically performed from histograms. As a proxy for a lower
case estimate in a ‘homogeneous’ material the 25th percentile lower case may be taken if data confidence is
high (say a large number of drill metres in that unit) or a 16th percentile if low. If data is poor, benchmarks
from other Pilbara designs or industry literature may be adopted. Lately, design values for defect strengths
in high confidence units have been taken from a reliable statistical database (Maldonado & Haile 2015;
Maldonado 2017).

5 Stability analyses and optimisation

The pit is domained into sectors of reasonably similar geotechnical characteristics. For each domain, stability
analyses are carried out at the batter scale using kinematic and limit equilibrium analyses, as required, using
the most realistic case only.
At the inter-ramp angle (IRA) scale, typically 2D limit equilibrium analyses are carried out except in large scale
complex environments where 3D numerical modelling may be adopted. Slopes are analysed to report a
Factor of Safety (FoS) based on the most realistic case and sensitivity analyses on individual lower case
elements, where these parameters are assessed as material to the design. The lowest stability result of the
lower case elements analysed is reported. Slopes are optimised to dual acceptance criteria for designated
slope risk options (Section 6).
From 2014–2017, indicative probabilities of failure were broadly inferred from typical industry FoS and PoF
relationships, based on published data in Wesseloo & Read (2009). From 2018 onwards, a normal distribution
of FoS was assumed. This normal distribution is constructed using the FoS results of the realistic case
(typically the 50th percentile) and the percentile estimate of the ‘controlling’ lower case. From this the
percentage of the curve at or below FoS 1 is taken as the PoF. This is illustrated in Figures 2 and 3.

Figure 2 Cumulative probability for lower/central case FoS 1.0/1.2, where lower case is 25th percentile

Figure 3 Normal probability density function for lower/central case FoS 1.0/1.2, where lower case is
25th percentile

418 Slope Stability 2020

 Safety and risk management

Minimum berm widths are also assessed at this stage using empirical relationships after the Modified Ritchie
criterion (Call 1992) and Ryan & Pryor (2000).

6 Acceptance criteria and slope options

Acceptance criteria have been set for:
• Batter stability (realistic case).
• Inter-ramp stability
○ Realistic case.
○ Lower case.
• Minimum berm width.
All must be satisfied in a particular domain. Some slopes are controlled by inter-ramp stability, whereas
others by batter/berm configuration.
A range of acceptance criteria are provided to the business for different slope risk options. These are termed
‘robust’, ‘balanced’ and ‘aggressive’, which have been defined appropriate to ‘critical infrastructure’, ‘typical
industry’ and ‘low risk’ mining environments respectively (Table 1). Note that in the context of ‘aggressive’,
‘low risk’ mining environments implies where the safety risk and the consequences of failure on the budgeted
mine plan are acceptably low, notwithstanding that the likelihood of failure is higher than other options.
At the inter-ramp scale, acceptance criteria were determined as follows. Firstly, central case factors of safety
for ‘critical infrastructure’, ‘typical industry’ and ‘low risk’ conditions, were taken from a review of typical
industry values published in Wesseloo & Read (2009). Secondly, where a published acceptance criteria in
Wesseloo & Read (2009) contained both a FoS and a corresponding PoF, these were plotted as data points
on a graph with FoS and PoF as x and y axes. Fit curves corresponding to low, moderate and high uncertainty
were plotted, again based data published in Wesseloo & Read (2009). Finally, the spread in FoS between
central case and lower case was taken from the typical uncertainty in the distribution of input parameters
for a low uncertainty environment. Low uncertainty was assumed given typical WAIO conditions where slope
heights are <150 m, repetition of the same geology / geotechnical units, failure mechanisms and broadly
similar strengths.

Table 1 Acceptance criteria for robust, balanced and aggressive slope options (H = batter height)

Acceptance IRA likely case IRA lower case Batter likely case Berm width
criteria (FoS) (FoS) (FoS, PoF) (m)
Robust >= 1.3 >= 1.1 >= 1.2, <= 15% 0.2H + 4.5
Balanced >= 1.2 >= 1.0 >= 1.1, <= 30% 0.2H + 4.5
Aggressive >= 1.15 >= 0.95 >= 1.0, <= 50% 0.17H + 3

An example summary of balanced case IRA results for a typical cutback are presented in Figure 4.
Once the final slope design has been reviewed, and revised if required, 3D slope solids representing each
domain and a set of batter berm configurations for each slope risk option are provided to mine planners for
high level pit design optimisation. Hazards (defined as sub-slope areas not meeting the acceptance criteria)
can be provided for risk owner consideration, where reasonably manageable and with reasonable benefit.
Such may be a local area within a domain with a low FoS that would otherwise control the domain, or allowing
single batters as hazards provided sufficient corresponding berm width ‘compensation’ is provided. The
design report also calls out specific design uncertainties and /or residual hazards. This is then made available
to the site based geotechnical engineers to design risk-based slope monitoring and reconciliation programs,
as required.

Slope Stability 2020 419

420
system: a simple pragmatic process for slope risk decisions
BHP Western Australia Iron Ore geotechnical open cut slope design

Figure 4 Example inter-ramp angle stability results for a typical balanced case (MLC = most likely (realistic case); LC = lower case; CY2/WA = material
controlling lower case)
A Haile et al.

Slope Stability 2020

 Safety and risk management

7 Simple risk assessment and value proposition

Risk assessment is performed at a high level in terms of likelihood and consequence.
For slopes controlled by batter/berm configuration, qualitative guidance is given on key risk aspects
impacting expected slope performance; for example, failure to achieve design on steeper slopes with
minimum berm widths can lead to an elevated safety risk profile, with the risk of ‘step-ins’ that sterilise ore.
Lately, berm reliability, reported in monthly compliance to design has been quoted to inform this risk.
For likelihood in slopes controlled by inter-ramp stability, this is assessed on the basis of PoF (as derived from
assumed statistical relationships between the central case and lower case FoS as described in Section 5).
Consequences for slopes controlled by inter-ramp stability are assessed using the in-house slope risk profile
tool. In the absence of incremental slope height analyses the tool factors the PoF against the exposed pit
depth, up to the maximum PoF at the full ‘daylight’ elevation. This is based on typical generic geotechnical
settings and failure mechanisms of the Pilbara. Run-out distance is based on an empirical best fit of models
by Finlay et al. (1999) and two models by Whittall (2015) namely the optimised model and his interpretation
of the stratified Fahrboschung angle for weathered weak rock. These assessments are based on a typically
representative generic 40 degree inter-ramp slope.
This is then factored by the representative strike length of the domain (it is assumed that strike length fails)
and the estimated PoF to return an indicative risk weighted impact in terms of expected metres squared of
mining area impacted by failure material by elevation. Risked impacts of each elevations are then summed
for all domains to show the cumulative risked impact area as a proportion of the total mining area for that
slope risk option. Figure 5 shows an example comparing balanced and aggressive results for a particular
cutback.

Figure 5 Example slope risk profile

These ‘simplistic’ analyses are considered suitable for the strategic level of the slope selection decisions;
however, more ‘optimised’ analyses should be conducted on site-specific parameters where these ‘generic’
relations are not considered suitable for strategic slope selection decisions. For example, impacts on critical
infrastructure such as a single access to the pit should be considered separately.
In terms of value proposition, since approximately 2017, mine planners have provided high level net present
value (NPV), or related tonnage estimates, for each slope risk option so that the relative value proposition
between them can be compared.

Slope Stability 2020 421

BHP Western Australia Iron Ore geotechnical open cut slope design A Haile et al.
system: a simple pragmatic process for slope risk decisions

8 Slope decisions
From 2014 to 2016, the operations based manager of production planning would select the slope option for
each domain. They would be supported at a meeting attended by key site, mine planning, geotechnical and
hydrological stakeholders. Each domain would be reviewed and particular issues raised such as indicative PoF,
hazards and particular consequences of failure, such as critical infrastructure (e.g. crusher or rail line), single or
dual pit accesses, interim or final wall or impacts of failure to achieve rockfall capacity or production targets.
Since 2017, the process is similar in that the geotechnical team provide the slope options; however, the
strategic mine planners now use these, with multi-disciplinary input, as the basis to make a single
recommendation for endorsement by the production manager. The production manager is the final decision
maker, being the single point of accountability for safety and production onsite. The main principle behind
the mine planners making a recommendation is that they can better inform production contingency planning
in the event of failure. In other words, they understand whether exposed ore is likely to be available
elsewhere in the event a slope were to fail and therefore the additional risk worth taking. An example of
slope selection presentation slide, for a single domain for typical cutback, is included as Figure 6. Table 2
shows the high grade feed difference for the cutback in Figures 4 and 5. The aggressive design allowed an
additional bench of pit depth to be mined compared to the robust / balanced designs.

Table 2 High grade feed difference for a cutback relative to aggressive design

Slope risk option Ore feed difference (%) Ore feed (tonnes)
Aggressive 0 0
Balanced -ve 3 -ve 702,000
Robust -ve 3 -ve 774,000

9 Results
A review of slope option selection in 2017 found that typically 70% of slope options selected were ‘aggressive’,
29% ‘balanced’ and 1% ‘robust’. On average, design slope angles have increased approximately 7 degrees since
pre-2013. As a rule of thumb, balanced slopes were steepened by 5 degrees and an additional 2 degrees was
achieved for the aggressive option. The uplift in the ‘balanced’ slope option is primarily attributed to the
introduction of the new design philosophy in 2014, with new common design principles incorporating:
• Realistic design principle including:
○ Central estimate strengths.
○ Reduction of ‘a’ value of 5 to 1 in the Snowden Linear Anisotropic Strength model (Mercer 2012).
• Inferred waviness at inter-ramp scale in bedding controlled slopes.
• Adoption of more double batters.
• The disturbance factor of D = 0.7 was constrained to the slope nearfield rather than across the
whole model.
In terms of outcomes, multi-batter slope failures have been few. Some sites have had issues in achieving
designs controlled by aggressive narrow berms and now tend not to select these.

422 Slope Stability 2020

Slope Stability 2020
Figure 6 Example domain slope risk profile communication slide

423
Safety and risk management
BHP Western Australia Iron Ore geotechnical open cut slope design A Haile et al.
system: a simple pragmatic process for slope risk decisions

10 Discussion
In some pre-2014 designs, the reported FoS was based on the 25th percentile due to uncertainty, or in other
cases the basis of a selected design value was not fully reported. The introduction of the realistic and lower
case design principles and clearer parameter derivation has made uncertainty transparent.
In addition, moving to consistent design principles the designs are altogether more consistent. This not only
offers a more realistic and consistent expression of risk but also in time generates a large, consistently
designed as-built dataset of pits, where design principles can be reconciled and improved (i.e. comparing
apples with apples).
Despite introducing lower case sensitivity analyses and associated acceptance criteria, where the lower case
criterion can control the slope selection, with the realistic case significantly exceeding the acceptance criteria,
overall designs have steepened. This is predominantly due to the various factors described in Section 9.
A further advantage of the lower case is that it offers flexibility to proceed to construction in areas of poor
data by drawing on lower cases estimated from benchmarks or literature.
Moving to estimates of PoF, although initially very simplistic, has been powerful in communicating relative
likelihood of failure between options. This has been more illuminating than quoting FoS alone, which some
parties, if not guided appropriately, can interpret as ‘this slope will not fail’. PoF also supports operational
staff tasked with designing monitoring and reconciliation programs. Such PoF estimates are now becoming
more refined (Section 7), however high probabilities of failure for IRA slopes controlled by rock mass have
not transpired into a proportional number of IRA failures to date. Investigation of this may bring more
refinement and confidence in the design process.
The different slope risk options ‘robust’, ‘balanced’ and ‘aggressive’, again support consistency and common
understanding across the business of both the ‘sensitivity’ of the slope to mining practice and the likelihood
of business disruption. However, the risk profile and necessary controls for successful implementation
requires careful communication, documentation, handover and regular refreshing to risk owners (managers
of production).
In practice, slopes can be implemented 2 to 10 years after taking slope option decisions. Records of slope
decisions are recorded; however, Production commitments to monitor and modify excavation practices as
required were not. Most managers taking slope decisions move before construction and commitments can
be lost. Understanding of the terms has improved, but it has taken several separate ore sterilisation events
for these lessons to be learnt.
Managers also need to be kept up to date with their risks. Having said that, despite being informed of leading
indicators on poor design compliance by geotechnical staff, this has not always led to successful resolution.
This is considered to be due to the inherent flexibility and optionality of the Pilbara iron ore mining
environment, such that sterilisation events typically do not immediately impact budgeted production. Finally,
incoming Managers need to be brought up to speed with the risk options, their risk profile and the controls
they own to be able to manage it appropriately.
None-the-less, the options of ‘robust’, ‘balanced’ and ‘aggressive’ are found to be fit-for-purpose to fulfil
their design intent of protecting critical infrastructure, enabling a typical mining environment and realising
more opportunity, respectively. In actuality, ‘robust’ has rarely been used; again due to the inherent flexibility
in mining operations and the remoteness of critical infrastructure from the influence of most pits.
We are still dependent on engineering judgement and compliance to design reconciliation to inform the
likelihood of ore sterilisation due to the cumulative loss of catch capacity in aggressive narrow berm designs.
A lot of value can be gained from tracking slope performance, e.g. berm reliability to inform corrective
actions, future slope decisions and investigation of slope over-performance, as described above.
In terms of assessing the consequences of slope failure, the risk profile tool is a simplistic representation of
the estimated run-out distance, based on a generic pre-failure 40 degree IRA, and assumes all the

424 Slope Stability 2020

 Safety and risk management

representative length of the domain fails and the impacted areas are not cleaned up (worst case scenario).
As these consequences are factored by likelihood (PoF), it does not represent the actual impact of a failure
event. However, for business planning purposes across BHP WAIO situation this would average out across all
deposits. Although simple, it does communicate indicative differences between risk options and can semi-
quantitatively highlight when to schedule availability of contingent ore (e.g. 2 benches above floor). It does
not capture narrow berm and specific ramp risks. Clearly, these should be called out separately.
Turning to slope decisions, in the early stages, without a value proposition, decision makers bemoaned a lack
of economic assessment (provision of NPV or ore tonnage differences) to support their decisions. Having
such information available is extremely valuable. Finally, the decision sits with the right person, they own the
safety risk and they are empowered to implement the controls. It is not a geotechnical accountability to pick
the risk profile for the company.

11 Conclusion
Since the implementation of the WAIO Geotechnical Design System in 2014:
The realistic design principle allows for
a. Transparency of known risks; and
b. Offers flexibility to proceed to construction in areas of poor data (acknowledging that with
more data the design could potentially be optimised further).
The risk profile and necessary controls requires careful communication, documentation, handover
and regular refreshing to existing and incoming risk owners.
Designs are consistent, offering a more reliable, transparent basis to take decisions, manage the
risk in operations and to reconcile and further optimise the system.
The system, particularly the new design principles, including the realistic design principle and
introduction of the ‘aggressive’ option, have steepened slope designs on average by about 7 degrees.
In execution, as a generalisation, inter-ramp stability has over-performed whilst aggressive narrow
berms in steeper slopes have under-performed and impacted on some of the anticipated savings.
Reconciliation offers the opportunity to highlight emerging risks and to refine reliability in the
design principles and methodology.
The risk profile tool, whilst simplistic, communicates the indicative difference between slope risk
options, which can semi-quantatively highlight when to schedule availability of contingency ore. It
does not however capture risks associated with narrow berms and specific ramps, which need to
be called out separately.
Decision makers struggle to take slope risk decisions without some indication of the economic
difference between risk options.
The risk owner, who is typically the Manager mining is formally engaged in collaborative discussions
with mine planning and geotechnical staff and has final say to endorse or reject the recommended
risk profile.

Acknowledgement
Rob Botha, BHP, Australia is thanked for his work on slope performance and significant design
non-compliance investigations. Michael Tucker, BHP, Australia is thanked for his early work on failure travel
distance and assessing the simplistic relationship to factor PoF based on depth exposed. Michael
Kostadinovski and Chaand Rajbhandari, BHP, Australia are thanked for their work on developing the most
recent slope risk option communication template. Chaand Rajbhandari is also thanked for her slope design
and risk profile work in the example presented in this paper. Finally, all members of the BHP WAIO

Slope Stability 2020 425

BHP Western Australia Iron Ore geotechnical open cut slope design A Haile et al.
system: a simple pragmatic process for slope risk decisions

geotechnical team are thanked for their contributions to the implementation and management of the system
over the past 5 years.

References
Call, RD 1992, ‘Slope Stability’, in HL Hartman (ed.), SME Mining Engineering Handbook, Society for Mining, Metallurgy & Exploration,
Englewood, pp. 881–896.
Finlay, PJ, Mostyn, GR & Fell, R 1999, ‘Landslide risk assessment: prediction of travel distance’, Canadian Geotechnical Journal, vol. 18,
no. 3, pp. 556–562.
Mercer, KG 2012, ‘The history and development of the anisotropic linear model: part 1’, Australian Centre for Geomechanics July
Newsletter, Australian Centre for Geomechanics, Perth, vol. 38, pp. 13–16.
Maldonado, A & Haile A, 2015, ‘Application of ANOVA and Tuckey-Cramer, statistical analysis to determine similarity of rock mass
strength properties across Banded Iron Formations of the Pilbara region in Western Australia’, Proceedings of the
International Symposium on Slope Stability in Open Pit Mining and Civil Engineering, South African Institute of Mining and
Metallurgy, Johannesburg, pp. 1–19.
Maldonado, A 2017, The basic friction angle and nominal roughness for Pilbara defects, BHP internal report.
Ryan, TM & Pryor, PR 2000, ‘Designing catch benches and interramp slopes’, in WA Hustrulid, MK Carter & DJA Van Zyl (eds), Slope
Stability in Surface Mining, SME Colorado, pp. 27–38.
Wesseloo, J & Read, J 2009, ‘Acceptance Criteria’, in J Read & P Stacey (eds), Guidelines for Open Pit Slope Design, CSIRO, Melbourne.
Whittall, JR 2015, Runout Exceedance Prediction for Open Pit Slope Failures, MSc Thesis, University of British Columbia, Vancouver.

426 Slope Stability 2020