Reconciliation –

Delivering on Promises
Bill Shaw, Andrew Weeks
Sia Khosrowshahi and Marcelo Godoy
Porto Alegre
6 November 2013
 Outline
Introduction - Mining Geology, Grade Control and Reconciliation
RECONCILIATION PRINCIPLES
A Question of Accuracy (And Reporting)
Defining Reconciliation
THE RECONCILIATION PROCESS
Ore Reserve (prediction) constraints on Reconciliation
Grade Control (prediction) constraints on Reconciliation
Mining (Production) Constraints on Reconciliation
Milling (Production) Constraints on Reconciliation
Basic Reconciliation Procedures
The Ideal Case
RECONCILIATION PRACTICE
A Consistent Framework for Factors
One Solution to the Problem of Uncertainty in Reconciliations
Other Types of Reconciliation
Outcomes of a Robust Reconciliation System
A NEW CASE STUDY
CONCLUSIONS.
2013 November 6 - Delivering on Promises
 Introduction
Mining Geology, Grade Control and
Reconciliation
 Read The JORC
Code
Transparency
Materiality
Competence
Definitions are in bold
Code is in normal typeface
Guidelines are in indented italics

2013 November 6 - Delivering on Promises

 A question of accuracy (and reporting)
JORC (2012, Table 1, Section 4) states:
“Where appropriate a statement of the relative accuracy
and confidence level in the Ore Reserve estimate using an
approach or procedure deemed appropriate by the
Competent Person. For example, the application of
statistical or geostatistical procedures to quantify the
relative accuracy of the reserve within stated confidence
limits, or, if such an approach is not deemed appropriate, a
qualitative discussion of the factors which could affect the
relative accuracy and confidence of the estimate. … …

2013 November 6 - Delivering on Promises

Additional requirements from the ASX
(mandatory 1 December, 2013)

“5.21.4 A comparison of the

mining entity’s mineral
resources and ore reserves
holdings against that from the
previous year on the following
basis:
by commodity type, including
the grade or quality; and
total ore reserves and total
mineral resources by
geographical area based on
the materiality of the mineral
resources and ore reserves
holdings to the mining entity,
…

2013 November 6 - Delivering on Promises

RECONCILIATION PRINCIPLES
Reconciliations change when the ore fed to
the mill changes
This can be due to
• Geology (lithology, alteration, weathering)
• Ore texture (hardness, grainsize, alteration)
• Sampling methods
• Interpretation methods
• Mining style
• Milling changes (equipment, throughput)
• Decision making (staff, management).

2013 November 6 - Delivering on Promises

 Basic Reconciliation Procedures
A simple scientific approach should enable a robust reconciliation
method to be quickly developed. The essential steps are:
1. Establish an audit trail for all data
2. Agree to report results routinely in a consistent format and ensure
that there are cross-functional reconciliation meetings in place to
discuss results and develop action plans
3. Tabulate the data
4. Report variations based on consistent volumes (bench by bench,
stope by stope) or periods (monthly, quarterly, annually)
5. Graph the variations (or factors) for each parameter to determine
trends
6. Analyse the differences and annotate the graphs to explain the
differences
7. Alter the input parameters systematically to reduce future
reconciliation differences.
2013 November 6 - Delivering on Promises
Data and Information

Wisdom

Knowledge

Information
Data

2013 November 6 - Delivering on Promises

The four stages for each component of the
reconciliation process
1. Data
2. Graphs
3. Annotation / Reporting
4. Decisions / Actions!

2013 November 6 - Delivering on Promises

 RECONCILIATION PRACTICE
A Consistent Framework for Factors
Consistent reconciliation principles
Always compare results against expectations
Always compare actual versus budget
Always compare measured with estimated
Always compare y axis to x axis
Always compare production and prediction.

2013 November 6 - Delivering on Promises

 A clear, consistent approach
Production
MILLING
STOCKPILES
MINING
ORE CONTROL
ORE RESERVES
MINERAL RESOURCES
GEOLOGY.

Prediction

2013 November 6 - Delivering on Promises

 F1, F2 and F3

Mine call factors and mill call factors have been

used in many mines without any clear systematic
definition
Harry Parker (2012) has provided a solution to
many of the grade control reconciliation problems,
since by his definitions …

2013 November 6 - Delivering on Promises

 Relationship between Factors
If:
short range model depletions GRADE CONTROL (PRODUCTION)
F1= ie F1=
long range model depletions ORE RESERVE (PREDICTION)
and
received at mill MILL (PRODUCTION)
F2= ie F2=
delivered to mill GRADE CONTROL (PREDICTION)
and
received at mill MILL (PRODUCTION)
F3= ie F3=
long range model depletions ORE RESERVE (PREDICTION)
then it is now obvious that F3 = F1 ∗ F2.

2013 November 6 - Delivering on Promises

Inputs:
Inputs to the Reconciliation Factors
Resource model
Exploration data
Original topography
Mining surface (start)
Mining surface (end)
Inputs:
Resource / Reserve GC model
Model
GC data
Inputs:
Grade Control
Ore block polygons
F1a model Stope design
Assigned grades Inputs:
Haulage records (source, destination,
tonnes, material type)
F1b Oreblock / Stope
Opening stocks
design
F1 Closing stocks Inputs:
Shift by shift crushing
Mining and records (Source, material
F2a
Stockpiling type, tonnes, moisture)
Opening and closing stocks

F2b Inputs:
Shift by shift processing
Crushing
records(tonnes, moisture,
head grade, recovery)
Opening and closing circuit
F2c stocks
Bullion measurements

F2 Processing

F3

2013 November 6 - Delivering on Promises

Example of an F1 Reconciliation

More ore tonnes at the expected grade

– check ore reserve interpretation?

More ore tonnes at lower grade

– check grade control dilution?

2013 November 6 - Delivering on Promises

 Understanding what the
reconciliation results mean
One solution to the problem of uncertainty
in reconciliations
If a model of an orebody has been created, the grade control, mining
and milling processes can be modelled. For example, the orebody
model can be sampled using various grade-control strategies and then
these notional grade-control samples can be used to predict outcomes
Such predictions can be compared to the complete model (the
exhaustive data set) from which the sample set is extracted
Such approaches are best based on conditional simulation modelling,
with the advantage that the transfer function, used to interrogate the
simulation realisations can be transparently and objectively
manipulated to seek different outcomes and provide a way for the
Grade Control and Mining processes to validate the Ore Reserve
predictions.

2013 November 6 - Delivering on Promises

 The Chain of Mining concept - improving
Ore Reserve predictions
The Chain of Mining approach defines the progressive
impact of:
• various grade control sampling regimes
• mining selectivity
• blasting
on the prediction of that part of the Mineral Resources that
are recoverable
This should be a significant consideration in converting
resources to Ore Reserves.

2013 November 6 - Delivering on Promises

 Summary of the Chain of Mining process
Reference GC Sample Errors Effective Selectivity Blasting

The CoM approach can be used to

validate the ore reserve model to check
recoverable tonnes and grade, and to
assess risk.
2013 November 6 - Delivering on Promises
 Effective Mining Selectivity (EMS)
The impact of all the contributing factors to the actual mining
selectivity can be determined on a case by case basis

This is important since differences can arise between the budget

tonnes and grade from an ore reserve and the actual recoveries
achieved by mining

The differences may be due to unrealistic parameters being assumed,

based on global SMU correction factors
Actual mining practice may
perform quite differently!

2013 November 6 - Delivering on Promises

The actual Effective Mining Selectivity
Impact of different mining sizes on ore block
definition (eg 5x5, 7.5x7.5, 10x10, 15x15).
Waste
HG placer
MG placer
HG sulphide
MG sulphide
LG sulphide

2013 November 6 - Delivering on Promises

Modelling the Effective Mining Selectivity
Predictions Mining decisions True reference
(tonnes/grades)
model

OK

ID2 This models the ore selection part of

the Chain of Mining process.
2013 November 6 - Delivering on Promises
 Modelling the impact of blasting
Model Lateral Vertical Result
displacement heave

This models the blasting part of

the Chain of Mining process.
2013 November 6 - Delivering on Promises
 Recap of the Chain of Mining process
Reference GC Sample Errors Effective Selectivity Blasting

2013 November 6 - Delivering on Promises

Other Types of Reconciliation
Example of a spatial reconciliation of
Grade Control vs Ore Reserves

Blue is where the model successfully predicted ore

Red is where the predicted ore was found to be low grade
White regions are where waste and ore were incorrectly
predicted by the ore reserve model.

2013 November 6 - Delivering on Promises

 The need for reconciliation
Reconciliations should be consistently monitored over time
Even a successful predictive approach can deteriorate due
to changes in geology, ore type, sampling procedures,
grade control methods, mining methods, milling controls,
personnel, etc
Lack of systematic reconciliation means that there are no
controls to monitor the predictions, and to moderate
expectations
This may result in non-optimal use of the resource,
pressure on the mining team, profit objectives not being
met and unhappy shareholders.
2013 November 6 - Delivering on Promises
 Outcomes of a
robust reconciliation system
Recognition of trends can provide insight into how the current
predictions may become realised during future production
It is useful to know that the mill is receiving the predicted ore at a
lower than expected grade, even while there is still uncertainty as to
whether this is due to problems with the:
• ore reserve (due to data, interpretation or estimation)
• grade control (due to similar errors plus ore loss and dilution)
• mining (due to deviations from the plan), and / or
• milling (due to sampling errors or losses)
Similarly it is useful to know that production is exceeding predictions
since this may mean the grade control process, the mine plan and the
revenues are all suboptimal.

2013 November 6 - Delivering on Promises

 Recommended approach to reconciliation
• Establish a robust reconciliation procedure

• Critically monitor both underestimation and

overestimation

• Identify problems

• Test and evaluate improvements

• Maintain clear, consistent documentation and

communication.

2013 November 6 - Delivering on Promises

 Advantages of a good reconciliation
process
Once problems have been highlighted solutions can be
considered. Typical examples are:
Problem Solution
Cannot achieve ore reserve Compare mapping and geological
model
Tonnage is too high Examine moisture content
Tonnage is too low Examine bulk density
Mill has less ore than mining Check stockpiles and weightometers
Mill has lower head grades Check circuit sampling and tailings.
We have been successfully using conditional simulation based methods for
10 years to evaluate ore reserves and to improve the objective repeatability of
ore selection for mining at the grade control stage.
2013 November 6 - Delivering on Promises
A Case Study from Mount Magnet
Gold Project, Western Australia
 Mount Magnet Gold Project

 Low grade operation

 Saturn Pit Reserves ~1.7g/t Au (at 30 June 2013)

 Shear-hosted, sub-vertical gold mineralisation

 Basalt and BIF host rocks, cross-cutting felsic dykes
 Gold mineralisation is highly variable and not visible

2013 November 6 - Delivering on Promises

 Simulation versus estimation
A 10% difference in the predicted grade of the ore blocks is noted
between these two models due to improved boundary resolution by
considering local data.
Simulation Model Kriged Model

2013 November 6 - Delivering on Promises

 Manual design of ore block boundaries
Commonly observed limitations with manual design of ore blocks
include subjectivity and inconsistency.

2013 November 6 - Delivering on Promises

Optimised versus manual ore blocks
At Mount Magnet, the resultant ore
block designs are very different using
the conditional simulation based
approach
In this example, the optimised ore
block boundaries yield a 26% increase
in recovered ounces compared to the
manual ore block design.

2013 November 6 - Delivering on Promises

 Results
Ore block design improvements have contributed to a 20% increase
in mill head grade.

2013 November 6 - Delivering on Promises

 An older case study
comparing Milled Grade and Predicted Grade
Golden Ridge
Predicted v Milled Grade

3.5

3 RC drilling &
OBO Implemented

2.5
Grade Au g/t

1.5

1
Supergene Transition Zone Primary
0.5

0
O Ja Ap Ju O Ja Ap
ct n- r-9 l-9 ct n- r-0
-9 99 9 -9 00
8 9 9 0

Predicted Grade Mill Reconciled Grade

2013 November 6 - Delivering on Promises

 Conclusions
A robust reconciliation system enables:
• the total mining operation to be seen in context
• major problems and sources of error to be identified
• both underestimation and overestimation to be critically
monitored
• improvements to be tested and evaluated
• reporting to management and communication to
shareholders to be clear and consistent.
Reconciling from the resource through to delivery of a
mineral product is the key to adding value during
development of a mining project.
2013 November 6 - Delivering on Promises
 Acknowledgements
Colleagues in writing this and other papers:
• Marcelo Godoy
• Sia Khosrowshahi
• Andrew Weeks
Many students on > 100 Grade Control workshops
João Felipe Costa and Rodrigo Peroni during the new
workshop on Reconciliation in Porto Alegre
16 new ‘disciples’ to this approach that will help with
change management in the mining industry to produce
better reconciliations between the Mill, Mining and Ore
Reserves and consequently improved estimates.
2013 November 6 - Delivering on Promises
 Obrigado.
November 2013

2013 November 6 - Delivering on Promises