Professional Documents
Culture Documents
net/publication/305744550
Article in Transactions of the Institution of Mining and Metallurgy, Section A: Mining Technology · July 2016
DOI: 10.1080/14749009.2016.1212510
CITATION READS
1 103
3 authors, including:
Kazuhiro Kawahata
5 PUBLICATIONS 38 CITATIONS
SEE PROFILE
All content following this page was uploaded by Kazuhiro Kawahata on 17 April 2018.
To cite this article: Kazuhiro Kawahata, Paul Schumacher & Kurt Criss (2016): Large-scale mine
production scheduling optimisation with mill blending constraints at Newmont’s Twin Creeks
Operation, Mining Technology, DOI: 10.1080/14749009.2016.1212510
Article views: 11
Download by: [T&F Office Locations], [Kazuhiro Kawahata] Date: 25 August 2016, At: 10:04
Large-scale mine production scheduling
optimisation with mill blending constraints at
Newmont’s Twin Creeks Operation
Kazuhiro Kawahata1*, Paul Schumacher1 and Kurt Criss2
This paper discusses the strategic mine planning process at Newmont’s Twin Creeks Operation.
A multi-period Mixed Integer Liner Programming (MILP) model is utilised to optimise a large scale,
complex operation with multiple open pits and underground mine sources under tight process blending
constraints with the option to stockpile. In addition to life-of-mine production scheduling optimisation
using MILP, the strategic planning process includes cost analysis, phase design validations and sensitivity
to constraints such as mining rates, mill blending or throughput. Through this planning approach, Twin
Creeks Operation is realising significant value from cost reductions and higher near term free cash flow
resulting in increases to life-of-mine net present value.
Keywords: Open Pit Mining, Production Scheduling, Optimisation
Seq 1
Seq 2
… Met type 1, Grade interval 1
Source i Increment k Met type 1, Grade interval 2
Seq j …
Sequence j etc.
Pton_mintd ≤ Xi,j,k,d
t
∑∑∑
i∈I j∈J k∈K
(1.2)
A t
≤ Pton_maxdt ∀ d ∈ 𝔻, t ∈ 𝕋
∑∑
+ n,k,d
n∈N k∈K
Mton_minti ≤ Xi,j,k,d
t
∑∑∑
j∈J k∈K d∈D
(1.3)
Zi,j,k,n
t
≤ Mton_maxit ∀ i ∈ 𝕀, t ∈ 𝕋
∑∑∑
+
j∈J k∈K n∈N
Xi,j,k,d
t
(Gi,j,k − Gmintd )
∑∑∑
i∈I j∈J k∈K (1.4) 3 Twin Creeks optimisation model
A t
(G − Gmin ) ≥ 0 t
∀ d ∈ 𝔻, t ∈ 𝕋
∑∑
�
+ n,k,d k d
n∈N k∈K
Objective function (1.1) maximises NPV. Because Newmont’s
Optimiser considers the life-of-mine time period, short-term
Xi,j,k,d
t
(Gi,j,k − Gmaxdt )
∑∑∑
goals are optimised without compromising long-term goals.
i∈I j∈J k∈K Constraints (1.2) and (1.3) enforce minimum and maxi-
Atn,k,d (Gk� − Gmaxdt ) ≤ 0 ∀ d ∈ 𝔻, t ∈ 𝕋 mum process and mining capacity constraints. Constraints
∑∑
+
n∈N k∈K (1.5) (1.4) and (1.5) enforce minimum and maximum process
blending constraints. These can be used for gold, SS, CO3 or
CORG. Constraints (1.6) ensure stockpile inventory balance
Ik,n
t
= Ik,n
t−1
Zi,j,k,n
t
Atn,k,d ∀ k ∈ 𝕂, n ∈ ℕ, t ∈ 𝕋 at each time period by tracking the previous time period’s
∑∑ ∑
+ −
i∈I j∈J n∈D stockpile ending inventory and the current period’s incom-
(1.6) ing and rehandled tons. Constraints (1.7) and (1.8) define
cumulative proportion variables and ensure material mined
does not exceed what is available in a deposit. Constraints
( )
Xi,j,k,d
t
Zi,j,k,n
t
∑∑∑ �
∑∑∑ �
+ (1.9) enforce each increment within a sequence is mined at
k∈K d∈D t � ≤t k∈K n∈N t � ≤t (1.7) the same rate as the overall sequence.
−P t
Resi,j,k = 0 ∀ i ∈ 𝕀, j ∈ 𝕁, t ∈ 𝕋 Constraints (1.10) and (1.11) ensure sequence j can-
∑
i,j
k∈K not be mined in time period t unless all its prerequisite
sequences j � ∈ 𝕁prev (j) are mined out at or before time period
Pi,j0 = 0, Pi,jt ≤ 1, Pi,jt ≤ Pi,jt+1 ∀ i ∈ 𝕀, j ∈ 𝕁, t ∈ 𝕋 (1.8) t. Constraints (1.12) through (1.16) restrict non-negativity
and (1.17) ensures binary.
Since many of processing facilities have relatively large
Xi,j,k,d
t
Zi,j,k,n
t
∑ ∑
+
fixed costs, the Optimiser has an option to handle fixed and
d∈D n∈N (1.9) variable costs separately. To keep a given facility open, mate-
= Pi,jt − Pi,jt−1 Resi,j,k ∀ i ∈ 𝕀, j ∈ 𝕁, k ∈ 𝕂, t ∈ 𝕋
( )
rials processed need to cover both fixed and variable costs,
but incremental tons only need to generate revenue sufficient
Pi,jt � − Yi,jt � ≥ 0 ∀ i ∈ 𝕀, j � ∈ 𝕁prev (j), t ∈ 𝕋 (1.10) to pay for variable costs. This allows the Optimiser to make
a more appropriate process shut down and start-up decisions.
Details are discussed in Hoerger et al. (1999).
Pi,jt − Yi,jt � ≤ 0 ∀ i ∈ 𝕀, j ∈ 𝕁, j � ∈ 𝕁prev (j), t ∈ 𝕋 (1.11)
Mining Technology 2016 3
Kawahata et al. Large-scale mine production scheduling optimisation at Newmont’s Twin Creeks Operation
(IBM 2014) on a 64-bit workstation with two Intel processors Cumulative Discounted CF Delta (vs. Base)
15%
Full Potential Programme
10%
To enhance operational excellence and deliver improved lev- 5%
els of operating cash flow, Newmont implemented a struc-
0%
tured and integrated programme called ‘Full Potential.’ At
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
Twin Creeks, the Full Potential programme commenced in
early 2014. Its goal is to sustainably improve operating cash
flow, with a specific focus on improving the first three-year 4 Theoretical mine rate analysis (discounted cash flow
cash flow without compromising life-of-mine NPV. delta to the existing mine plan)
Newmont’s main focus of strategic planning under Full
Potential is to identify and provide a high-level estimate of
benefits from any fundamental mine planning and processing Yearly Mined Tons (New Mine Plan vs. Base Plan)
70
changes that can be made to improve performance. Using
Millions
Pit 1, Phase 1
the Optimiser model as a platform, this work included phase 60 Pit 1, Phase 2
Pit 1, Phase 3
design evaluations and validation, production scheduling
50 Pit 1, Phase 4
optimisation, and investigations of potential additional ore Pit 2, Phase 1
Mined Tons
sources. Additionally, the team evaluated alternative pro- 40 Pit 2, Phase 2
cessing methods, mill blending strategies and variations in Pit 2, Phase 3
30
offsite-sourced concentrate additions to help unlock the value Base Mine Rate
Regional impact Clark, L. and Dagdelen, K. 2007. Finding added value at Newmont, Mining
Magazine, 196, (4), 29–35.
Due to interdependencies between sites in Newmont’s Nevada Dagdelen, K. 1996. Formulation of open pit scheduling problem including
sequencing as MILP, Internal Report. Golden, CO, Mining Engineering
operations, the site planning process should also consider broader Department, Colorado School of Mines.
regional impacts to make sure improvements at a site level do Dagdelen, K. and Kawahata, K. 2007. Opportunities in multi-mine planning
not adversely affect results at the regional level. With help from through large scale mixed integer linear programming optimization, 33rd
the regional planning group, Twin Creeks’ mine plan changes International Symposium on Application of Computers and Operations
and various blending strategies are validated by running the full Research in the Mineral Industry (APCOM), Santiago, Chile, 337–342.
Dagdelen, K. and Kawahata, K. 2008. Value creation through strategic mine
Nevada-wide Optimiser model that includes the Carlin opera- planning and cutoff grade optimization, Mining Engineering. Society for
tions. This process must confirm that additional value from Twin Mining, Metallurgy and Exploration (SME), 60, (1), 39–45.
Creeks’ plan improvements is realisable in the regional portfolio. Goodfellow, R. and Dimitrakopoulos, R. 2013. Algorithmic integration of
geological uncertainty in pushback designs for complex multiprocess open
pit mines, Mining Technology, 122, (2), 67–77.
Conclusion Goodfellow, R. and Dimitrakopoulos, R. 2015. Global optimization of open pit
mining complexes with uncertainty, Applied Soft Computing, 40, 292–304.
The Twin Creeks team has successfully utilised Newmont’s doi:http://dx.doi.org/10.1016/j.asoc.2015.11.038.
MILP-based Optimiser in Full Potential strategic planning. Hoerger, S., Bachmann, J., Criss, K. and Shortridge, E. 1999. Long term
mine and process scheduling at Newmont’s Nevada operations, 28th
Scenario analysis performed using this type of life-of-mine International Symposium on Application of Computers and Operations
optimisation model allowed strategic changes to be evaluated Research in the Mineral Industry (APCOM), 739–747.
in terms of their impact on the whole Nevada-wide system by Huang, Z., Cai, W. and Banfield, A. F. 2009. A new short- and medium-term
considering all future implications simultaneously. Strength production planning tool – Minesight® Schedule Optimizer (MSSO), SME
of this approach also came from being able to evaluate more Preprint 09-079, Denver, Colorado.
IBM. 2014. IBM ILOG CPLEX Optimizer. www.ibm.com.
than 50 scenarios quickly and then selecting only a few viable Kawahata, K., Schumacher, P. and Hufford, R. 2013. Mine production
scenarios for more detailed work. This optimisation process scheduling optimization at Newmont’s Twin Creeks Mine, Mining
provided more confidence in the resulting production sched- Engineering. Society for Mining, Metallurgy and Exploration (SME), 65,
ule and helped planners to better understand plan sensitivity. (6), 49–54.
Kuchta, M., Newman, A. and Topal, E. 2004. Implementing a production
In addition to the strategy work stream discussed here, schedule at LKAB’s Kiruna Mine. Interfaces, 34, (2), 124–134.
application of Full Potential to the performance improvement Maptek. 2009. Chronos scheduler, Sydney, Maptek Software.
work stream is identifying several operational opportunities. Minemax. 2012. Minemax scheduler strategic mine schedule optimisation
Through these combined efforts, Twin Creeks has so far white paper, Perth, Minemax.
improved the life-of-mine NPV by $120M and projects reduc- Montiel, L. and Dimitrakopoulos, R. 2015. Optimizing mining complexes with
multiple processing and transportation alternatives: An uncertainty-based
ing operating costs by up to $50M over the next three years. approach, European Journal of Operational Research, 247, (1), 166–178.
doi:http://dx.doi.org/10.1016/j.ejor.2015.05.002.
Montiel, L., Dimitrakopoulos, R. and Kawahata, K. 2015. Globally optimising
Acknowledgements open-pit and underground mining operations under geological uncertainty,
Mining Technology (TIMM A), 125, (1), 2–14. doi:http://dx.doi.
Authors thank Twin Creeks Site Leadership Team and org/10.1179/1743286315Y.27.
Newmont. 2014. 2013 Newmont Annual Report and Form 10-K, Greenwood
Newmont North America Leadership Team for their support Village, Colorado, Newmont Mining Corporation.
and willingness to publish this work. We also thank Dr. Terry Ramazan, S. and Dimitrakopoulos, R. 2004. Traditional and new MIP models
Bush for his help and Dr. Conor Meagher for reviewing this for production scheduling with in-situ grade variability, International
paper. Journal of Surface Mining, Reclamation and Environment, 18, (2), 85–98.
Seymour, F. 1996. Pit-limit parameterization from modified three-dimensional
Lerches-Grossmann algorithm, SME Transactions, 298, 1860–1864.
Stone, P., Froyland, G., Menabde, M,Law, B., Rasyar, R. and Monkhouse,
References P. 2004. “Blasor – Blended iron-ore mine planning optimisation at Yandi”.
Orebody Modelling and Strategic Mine Planning. In: Proceedings of the
Asad, M. W. A. and Dimitrakopoulos, R. 2012. Performance evaluation of a International Symposium, 285–288, Preprint, Carlton, Victoria: AIMM.
new stochastic network flow approach to optimal open pit mine design- Urbaez, E. and Dagdelen, K. 1999. Implementation of linear programming
application at a gold mine, Journal of the Southern African Institute of model for optimum open pit production scheduling problem, SME Annual
Mining and Metallurgy, 112, (7), 649–655. Meeting Preprint, Denver, Colorado, 99–108.
Askari-Nasab, H., Awuah-Offei, K. and Eivazy, H. 2010. Large-scale open Whittle, J. 2009. The global optimizer works – what next?. Advances in
pit production scheduling using mixed integer linear programming, Orebody Modelling and Strategic Mine Planning I. The Australasian
International Journal of Mining and Mineral Engineering, 2, (3), 185–214. Institute of Mining and Metallurgy, Melbourne, 3–5.
Mining Technology 2016 5